Law and the Limits of Reason
This page intentionally left blank
Law and the Limits of Reason Adrian Vermeule
1
1...
83 downloads
1408 Views
1MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Law and the Limits of Reason
This page intentionally left blank
Law and the Limits of Reason Adrian Vermeule
1
1 Oxford University Press, Inc., publishes works that further Oxford University’s objective of excellence in research, scholarship, and education. Oxford New York Auckland Cape Town Dar es Salaam Hong Kong Karachi Kuala Lumpur Mexico City Nairobi New Delhi Shanghai Taipei Toronto
Madrid
Melbourne
With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore South Korea Switzerland Thailand Turkey Ukraine Vietnam
Copyright © 2009 by Oxford University Press, Inc. Published by Oxford University Press, Inc. 198 Madison Avenue, New York, New York 10016 Oxford is a registered trademark of Oxford University Press Oxford University Press is a registered trademark of Oxford University Press, Inc. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press, Inc. _____________________________________________ Library of Congress Cataloging-in-Publication Data Vermeule, Adrian, 1968Law and the limits of reason / Adrian Vermeule. p. cm. Includes bibliographical references and index. ISBN 978-0-19-538376-8 ((hardback) : alk. paper) 1. Law—Methodology. 2. Law—United States—Methodology. I. Title. K212. V47 2008 340’.1—dc22 2008032379 _____________________________________________
1 2 3 4 5 6 7 8 9 Printed in the United States of America on acid-free paper Note to Readers This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is based upon sources believed to be accurate and reliable and is intended to be current as of the time it was written. It is sold with the understanding that the publisher is not engaged in rendering legal, accounting, or other professional services. If legal advice or other expert assistance is required, the services of a competent professional person should be sought. Also, to confirm that the information has not been affected or changed by recent developments, traditional legal research techniques should be used, including checking primary sources where appropriate. (Based on the Declaration of Principles jointly adopted by a Committee of the American Bar Association and a Committee of Publishers and Associations.)
You may order this or any other Oxford University Press publication by visiting the Oxford University Press website at www.oup.com
For My Family
This page intentionally left blank
Contents
Introduction: The Limits of Reason in Legal Theory . . . . . . . . . . . . . . . .
1
1. Many-Minds Arguments . . . . . . . . . . . . . . . . . . . . . . . . . .
25
2. The Constitutional Common Law: Information Aggregation . . . . . . . . . . . . . . . . . . . . . . . . . .
57
3. The Constitutional Common Law: Evolution . . . . . . . . .
97
4. Justices and Company . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 5. Unintended Consequences and Constitutional Amendments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
Conclusion: The Codified Constitution . . . . . . . . . . . . . . . . . . . . . . . . . . 187 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195 Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
This page intentionally left blank
Introduction
The Limits of Reason in Legal Theory “As to ‘common reason,’ or the reason of the majority of the people who use their reason about the matter, whose reason is it most to be apprehended should run counter to it? That of many hundred men chosen the greater part of them by the people . . . or that of four men [i.e. judges], appointed by the Crown[?]” — Jeremy Bentham1
R
eason is a scarce commodity; the legal system must simultaneously encourage its production and deploy it for maximum advantage. Reason is limited by the intrinsic costs of collecting information, by the difficulty of knowing the value of information one has not yet collected, by the processing capacity of the brain, by errors arising from cognitive heuristics and biases, by the distorting force of self-interest, and by emotional distortions (although it is also true that heuristics may be cheap and accurate on average, self-interest can motivate sharpened cognition, and emotions are sometimes necessary for reason to function). Given the 1
Jeremy Bentham, A Comment on the Commentaries 154 (Charles W. Everett ed., 1928).
1
2
L L R
scarcity of reason, how should lawmaking authority be allocated among judges, legislators, and executive officials, and how should legal institutions be designed? In what follows, my largest ambition is to answer these questions.
Epistemic Legalism: A Definition and Critique I will break down the overall project into two subsidiary aims, the first critical, the second constructive. The critical aim is to evaluate a family of views that I shall call epistemic legalism. Legalists prefer judicial lawmaking to lawmaking by other institutions; epistemic legalists claim that the limits of reason support judicial lawmaking over lawmaking by other institutions—legislatures, executive officials, and administrative agencies. Obviously, this is not a fully specified claim; for one thing, we must ask “judicial lawmaking of what kind?” Judge-made common law may be compared to legislation or administrative lawmaking, judge-made constitutional law may be compared to legislation or constitutional amendment, and so on. Different ways of specifying epistemic legalism will thus yield different views, addressing somewhat different issues. However, what unites the various strands of epistemic legalism is a view that the limits of reason support a large and perhaps even predominant role for judicial lawmaking. Epistemic legalism thus includes some of the most striking and important claims in legal theory, such as the claims of Edmund Burke and F.A. von Hayek that the limits of reason entail the superiority of tradition, custom and common law over legislation. It also includes arguments by modern legal scholars for “common-law constitutionalism”2—that is, constitutional law made by judges through the accretion of precedent—over constitutional lawmaking through formal amendments or legislative and executive action. In general, the claims of epistemic legalism are central to legal theory, intellectually ambitious, and wrong, or so my negative thesis claims. There is no logical connection between the limits of reason, on the one hand, and the superiority of common law or of judge-made constitutional law, on the other. Nor, apart from logic, is there even a general and 2
I do not use “constitutional common law” in Henry Monaghan’s sense of constitutionally inspired doctrine that might be overridden by legislation. See Henry Paul Monaghan, The Supreme Court, 1974 Term—Foreword: Constitutional Common Law, 89 Harv. L. Rev. 1 (1975). Rather, I use the term to denote constitutional rules, superior to ordinary legislation, that are elaborated by judges through precedent-based reasoning. Synonyms might include “judgemade constitutional law” or “common-law constitutional exegesis.”
T L R L T
3
robust causal connection between these two things. Mechanisms that promote epistemic success—mechanisms of information aggregation, institutional and legal evolution, and deliberation—do not hook up to the preference for judicial lawmaking that is the defining feature of epistemic legalism, and which supplies its tonality. The argument to judicial lawmaking from the limits of reason outruns the logical, causal, and evidentiary support. The relatively small number of judges on relevant courts, their limited informational base, their professional homogeneity, and their generalist rather than specialized skills, all ensure that judicial reason is itself sharply limited. Contra Burke, Hayek, and other theorists I shall examine, these deficits are not offset by the judges’ ability to call upon the purported collective wisdom of past judges or of social traditions and customs. Under certain conditions, the extension of judicial decisionmaking over time is a demerit rather than an advantage; where it is an advantage, it is one that legislatures can also exploit and thus not an advantage that is unique to the judges; and, most importantly, legislatures possess epistemic advantages by virtue of their diversity, their tools for gathering and processing information, and the sheer number of their members. In general, epistemic legalists frequently elide two very different claims: that many judicial heads are better than one and that judicial heads are better than legislative and executive ones. Even where the former claim is true, it does nothing to establish the latter claim. And under a broad range of conditions, legislatures and executive bodies do indeed possess superior epistemic capacities, or so I argue throughout.
Levels of Rationalism and Institutional Comparisons This critique of epistemic legalism rests on a claim about the comparative epistemic competence of institutions; that claim has complex connections to rationalism, in different senses. Critics of the common law, and more recently critics of the constitutional common law, are often charged with excessive rationalism. On this view, legislation and constitutional amendment are ambitious instruments of centralized social planning (as Hayekians charge) or result from excessive confidence in the powers of theoretical reason (as Burkeans charge). One of my central claims, however, is that these charges fail. They rest on comparisons between one mind and many minds that are irrelevant to evaluating the epistemic capacities of many-headed legislatures; alternatively, or in addition, they fall into a kind of nirvana fallacy by overlooking that the same institutional features
4
L L R
relied upon to impeach legislative rationalism also impeach the rationality of judicial lawmaking, even more strongly. In this analysis, critics of the (constitutional) common law should not let themselves be depicted as committed to unconstrained rationalism, while celebrants of the common law claim the mantle of respect for reason’s inherent limitations. My suggestion is that praise for the (constitutional) common law rests on an incomplete or asymmetrical second-order rationalism: a form of nirvana fallacy that uses rational argument to impeach the first-order rationalism of legislative institutions and processes, but that fails to apply its analysis evenhandedly to judicial institutions. This sort of view generates an idealized picture or best-case scenario for judicial lawmaking and compares it to a worst-case scenario for legislative lawmaking. In an evenhanded comparison, the epistemic superiority of legislative lawmaking emerges under a broad range of conditions. Epistemic legalism, in other words, represents an unhappy and untenable way station between unbridled first-order rationalism, on the one hand, and a thoroughgoing and evenhanded second-order rationalism, on the other. Once we have reached the stage of institutionally consistent second-order rationalism, the links between epistemic legalism and the limits of reason fall apart. In this sense, my project is an effort to reclaim the limits of reason from the celebrants of the common law, who have illegitimately appropriated them for their institutional cause.
The Codified Constitution So much for the critique of epistemic legalism. My constructive program is to propose a legal regime that I will call the codified constitution. I attempt to show that, under robust conditions, the limits of reason affirmatively support a larger role for lawmaking by legislatures and executive officials than is allowed under epistemic legalism—emphatically including constitutional lawmaking. Precisely because of the limits of human reason, large modern legislatures, with their numerous memberships, professional and ideological diversity, complex internal structures for processing information, and abundant informational resources, are epistemically formidable institutions. The codified constitution means that statutes and constitutional amendments, rather than judicial precedents in the common-law style, will do the bulk of the work in filling constitutional gaps, clarifying constitutional ambiguities, and updating constitutional law under changing circumstances.
T L R L T
5
Structure To orient the book’s thesis, claims, and themes, I will briefly lay out the structure of the following chapters. Chapter 1 introduces the central epistemic mechanisms by discussing “many-minds” arguments in legal theory and in neighboring sectors of political theory, political science, psychology, and economics. The key mechanisms involve information aggregation, institutional evolution, and deliberation; I give special scrutiny to recent claims for the importance of the Condorcet Jury Theorem and the “wisdom of crowds,” for the most part with a jaundiced eye. The phrase “many minds” stems from pioneering work by Cass Sunstein,3 but my conclusions are very different from Sunstein’s. I am skeptical that manyminds arguments are general or robust, especially skeptical that they support a robust role for judicial lawmaking, and I deny outright that they support common-law constitutionalism. Chapters 2 and 3 focus on Burkean traditionalism in American legal theory, in particular constitutional theory, and on attempts by constitutional theorists to appropriate Burke by elaborating an epistemic rationale for his views. Burke’s account of tradition and its relationship to the common law is evocative but ill-specified, so it is necessary to consider both an informational interpretation of Burke centering on the Jury Theorem, and an evolutionary interpretation centering on invisible-hand mechanisms alleged to make the common law efficient, or informationladen, or otherwise socially desirable or useful. The informational interpretation of Burke is taken up in Chapter 2, and the evolutionary one in Chapter 3, which also considers Hayek’s evolutionary account of the common law as a spontaneous order. In both chapters, I question the internal logic of the mechanisms said to give epistemic power to the common law generally and to constitutional common law in particular. Here a central claim is that there is a Burkean paradox: to rely on precedent or tradition or custom because it embodies the contributions of many minds is to make no fresh contribution to the collective wisdom; conversely, participants in the line of precedent or tradition or custom make an epistemic contribution only when, and because, they form independent views rather than deferring to tradition. Where actors defer to the information of past others, as the Burkean position would have them do, the result is a low-value “information cascade” rather than 3
Cass R. Sunstein, Infotopia: How Many Minds Produce Knowledge (2006); see also Cass R. Sunstein, A Constitution of Many Minds (forthcoming 2009).
6
L L R
collective wisdom. Burkeans may diminish the force of the paradox by reducing their reliance upon tradition, but then the claimed epistemic benefits of tradition are diluted as well; as the decisionmaking strategy becomes more distinctively Burkean, the paradox becomes stronger. Furthermore, and crucially, I argue that the epistemic mechanisms underpinning Burkean legal theory cannot successfully be transposed to the constitutional common law. The transposition rests on a fallacy: even if many minds are better than one, so that a single judge does better by relying on common-law precedent than on his own unaided reason, it does not at all follow that judicial lawmaking is superior to legislative lawmaking. The latter is a many-to-many comparison rather than a one-tomany comparison. When the relevant comparisons are performed evenhandedly across institutions, the conclusions change. The many current minds of legislators, aided by the epistemic resources and complex information-processing structures of modern legislatures, are epistemically superior to the collective wisdom of a line of judicial precedents. Chapter 4 changes tack by examining the design of legal institutions, as opposed to the allocation of tasks across institutions whose design is taken as fixed. Epistemic considerations, I argue, call for professional diversity in the judiciary, specifically on the United States Supreme Court: there should be at least some justices who are not lawyers. I pursue here both a substantive and a methodological point. Substantively, when the limits of reason are taken into account, the professional monopoly of judicial positions by lawyers is seen to be undesirable, contrary to a central sociological assumption of epistemic legalism. Methodologically, I hope to demonstrate an epistemic argument that survives all the grounds for skepticism laid out in Chapter 1. My primary quarrel is not with the epistemic mechanisms themselves, but with their judge-centered application in modern legal theory. Chapter 5 examines the constitutional common law from another angle, comparing it, not to statutes, but to constitutional amendments. I explore the epistemic benefits and costs of making constitutional law through positive legislatively enacted instruments, rather than judicial interpretation. Analytically, I also shift the focus from many-minds arguments to arguments based on bad unintended consequences. The limits of reason include limited human foresight; epistemic legalists claim that statutes and amendments routinely bring about unintended and possibly perverse consequences, when they accomplish anything at all. On these grounds, common-law constitutionalists have denigrated the process of constitutional amendment in the American legal system, claiming that
T L R L T
7
amendments are either systematically futile or systematically produce perverse or harmful consequences. I argue that the futility claim is false and that the perversity claim rests on a nirvana fallacy—a false comparison between the best-case scenario for common-law constitutionalism and the worst-case scenario for constitutional amendment. Judge-made law is also rife with unintended consequences; the inability of adjudication to take a systemic view of tradeoffs and interrelated variables exacerbates that risk. In this light, I attempt to state the conditions under which the process of constitutional amendment produces law that is superior, on epistemic grounds, to the processes of common-law constitutionalism. How do these proposals interact? In principle, the tasks of institutional design and the allocation of lawmaking authority should be pursued simultaneously. Conclusions on either margin should be adjusted until a kind of reflective equilibrium arises, in which institutions are designed in an epistemically optimal manner, given the allocation of lawmaking authority, and in which authority is allocated in an epistemically optimal manner, given the institutional design. In reality, however, one is constrained to discuss these issues piecemeal, seriatim, and in a partial rather than comprehensive fashion. At appropriate points in the discussion, however, I examine interactions among the various proposals. In general, the proposals to reallocate lawmaking authority from courts to legislatures on epistemic grounds (chapters 2, 3 and 5) and the proposal to diversify the judiciary on professional lines (Chapter 4) are partial substitutes: to the extent that one is pursued, the other becomes less important, but the overall optimum would plausibly be a combination of both reforms. So much for the main line of the argument; I will now state the major themes and make explicit some assumptions and limitations of the inquiry.
“Epistemic” Claims Defined A fundamental limitation, which should be kept in mind throughout, is that I examine only epistemic claims about the questions I will discuss. By “epistemic,” I mean arguments according to which legal rules and institutions, including allocations of decisionmaking authority across institutions, can do better or worse at tracking the truth.4 Domains where there are no right answers, or no truth to be tracked, simply lie outside my area of concern. 4
For “truth-tracking” as the nub of an epistemic argument, see Robert E. Goodin, Reflective Democracy (2003).
8
L L R
What truth exactly is being tracked will, of course, vary with context; it need not be factual, but can also be causal, legal (on standard accounts of jurisprudence), or moral (on standard accounts of metaethics). In general, nothing in the views I will advance here depends on the content of the relevant truths, although the institutional analysis will of course have to take into account which group of decisionmakers is best positioned to discern which set of truths. Furthermore, as I explain in Chapter 1, the relevant truths need not be exogenous in the sense that they are independent of the preferences of a decisionmaking group; they can be truths about what those preferences are. A common mistake is to think that epistemic arguments can only be evaluated, pro or con, if we know what the relevant truths actually are. But real-world decisionmakers and institutional designers can and often do evaluate epistemic arguments about various legal rules or institutional arrangements without knowing the underlying truths. Consider that it is perfectly ordinary, indeed inescapable, to ask whether, for example, an adversarial or inquisitorial system for producing evidence in criminal cases is better at tracking the truth, even when we do not know which accused criminals are in fact guilty and which ones are not. In general, we can and do evaluate the epistemic capacities of legal institutions indirectly, by looking at their inputs and processes rather than their outputs alone. And we do so with fair success.
Truth, Not Wisdom I will focus on truth, not on wisdom. “Wisdom” is usually understood as something different than, and in some sense higher or richer than, epistemic competence in the sense of accuracy or capacity to track the truth. I will deliberately bracket those higher mysteries. A focus on epistemic accuracy is simplistic, but this very simplicity is also its major advantage, as it makes tractable an inquiry that would otherwise risk falling into a philosophical morass. And the loss of conceptual richness is not large, at least for the questions I will examine. Few doubt that truth-tracking or epistemic accuracy is part of wisdom, if not the whole of it. The issues of institutional design and the allocation of lawmaking power across institutions that I discuss will rarely be sensitive to fine analytic distinctions between truth-tracking capacity, wisdom, and other epistemic or partially epistemic virtues. Finally, standard treatments of “the wisdom of crowds” use wisdom essentially as an impressive synonym for
T L R L T
9
accuracy.5 As my aim is to explore the limits and implications of those treatments, I will follow suit.
Epistemic and Nonepistemic Claims and Tradeoffs Common-law decisionmaking and common-law constitutionalism have also been defended on nonepistemic grounds, such as a preference for gradual over rapid change.6 I do not speak to these questions, so my conclusions should be understood as ceteris paribus claims and qualified accordingly. In a broader treatment, one would want to consider epistemic and nonepistemic questions in a systemic overview, allocating tasks to legal institutions, and designing those institutions, with a view to satisfying some combination of epistemic and nonepistemic aims. On a full consequentialist analysis, that is, the epistemic quality of institutional decisions is not something to be maximized, but rather optimized. Because maximizing epistemic quality typically counsels that more heads are better than one, and that cognitive diversity within the decisionmaking group is desirable, epistemic considerations tend to yield larger decisionmaking groups whose members have fewer common premises and higher costs of communication. Thus epistemic considerations will systematically raise the direct costs and opportunity costs of decisionmaking, and these costs must be traded off against the epistemic benefits. Moreover, in some domains, there is no truth to be tracked, so epistemic arguments get no purchase in the first place. I exclude nonepistemic issues and domains partly because I have addressed many of those issues elsewhere, insofar as they bear on the allocation of lawmaking and law-interpreting authority among legislatures, courts, and the executive;7 partly for the sake of focus and thematic unity; but most of all for the substantive reason that epistemic claims take pride of place in the views I shall critique. Burkeans, Hayekians, and other defenders of common-law processes and of common-law constitutionalism are rarely content to limit themselves to nonepistemic arguments. The epistemic ones hold the intellectual high ground; a foundation in truth seems a more secure and widely appealing basis for a given view than (say) a preference for a certain rate of change, which might appear 5 6 7
See, e.g., James Surowiecki, The Wisdom of Crowds (2004). See David L. Shapiro, Continuity and Change in Statutory Interpretation, 67 N.Y.U. L. Rev. 921, 941–50 (1992); Thomas W. Merrill, Bork v. Burke, 19 Harv. J.L. & Pub. Pol’y 509, 511–13 (1996). See Adrian Vermeule, Judging Under Uncertainty: An Institutional Theory of Legal Interpretation (2006).
10
L L R
idiosyncratic or sectarian. If its epistemic struts are flawed, the whole edifice of the common-law approach to constitutionalism will become shaky. While my claims are limited, the epistemic issues I will examine are crucial terrain for legal and political theory.
The Limits of Reason: Two Faces Epistemic legalism connects law with the limits of reason; but what exactly does epistemic legalism claim, or mean? I break down epistemic legalism into two components: arguments based on the virtues of many minds, and arguments based on the fear of unintended consequences. Both sets of arguments are associated especially with Burke and Hayek, who emphasize respect for tradition, custom, precedent, and evolved legal rules, norms, and institutions. These sources, the claim runs, help to overcome the limits of reason by contributing the information and perspectives of many minds that have considered similar problems over time. Legislative lawmaking, by contrast, is depicted as a form of excessive or overly ambitious rationalism that fails to respect the limits of reason. This last claim, about the limits of legislative action, also rests upon fear of unintended consequences. A broad cadre of academic legal theorists, especially constitutional theorists, are suspicious of legislative action and implicitly view the common law, and by extension the constitutional common law, as an organic unity that legislative action will disrupt. These theorists emphasize the limits of human foresight and the fear that attempts by legislators to improve constitutional law or raise social welfare through intentional action will result in futility (at best) or harmful blunders and perverse effects (at worst). Chapter 5 examines these arguments in their purest and legally most consequential form: the choice between commonlaw constitutionalism, on the one hand, and constitutional amendment, on the other, as alternative means for updating constitutional law over time. The distinction between the many-minds strand of epistemic legalism and the unintended-consequences strand is largely a matter of emphasis. Of course the two types of claims are intimately related. Hayek and his followers, for example, warn that legislation will produce bad unintended consequences, while they praise the putative evolutionary processes of the common law as producing a spontaneous order that coordinates social expectations. What links the two strands is their role in the overall view I have dubbed epistemic legalism. Arguments that invoke many minds to praise the common law and the constitutional common
T L R L T
11
law underscore the benefits of epistemic legalism, while arguments that emphasize the risk of unintended consequences from legislative action underscore the costs of the alternatives. Epistemic legalism is thus a Janusfaced view.
The Epistemic Superiority of Legislatures As against epistemic legalism, I defend the codified constitution, which results from the epistemic superiority of legislatures to courts under a broad range of conditions, especially in matters of constitutional law. In the comparison at the heart of the book—between current legislatures and current courts interpreting vague or ambiguous provisions of written constitutions—legislatures enjoy several major epistemic advantages. The first, often overlooked, is sheer numerosity. There are many more legislators in a typical national legislature than there are judges on a typical high constitutional court; although adding together all judges on all courts in the national judicial system may tip the numbers the other way, that extended group of judges is strictly notional, whereas national legislators actually work together as part of an integrated institution that votes upon and decides common questions. As we will see, the numerosity of modern legislatures is an important epistemic resource; in matters epistemic, so long as a minimum threshold of competence is met, numbers tell. Second, legislatures are more representative than courts, and representation produces knowledge. As seen in Bentham’s quotation with which I began, the relationship of representation gives legislators information about local conditions and social judgments and preferences that judges cannot hope to match. Moreover, modern legislatures possess powerful institutional machinery for generating information, both in their own right and through delegation to expert executive officials. Although representation can introduce political distortions that reduce epistemic competence, we will see that judges are not immune to political distortions either, and that judicial insulation from day-to-day politics makes judges especially prone to err when they attempt to adapt constitutional law to changing circumstances. Third, and crucially, a typical modern legislature is more diverse than a typical modern judiciary, especially along professional lines. Professional diversity reduces groupthink—the positive correlation of biases within decisionmaking groups—and is thus an important source of epistemic strength. Traditional legal and political theory focus myopically upon the judges’
12
L L R
expertise—their presumed epistemic competence—overlooking that competence is but one determinant of group epistemic performance, along with diversity and numbers. A more diverse and more numerous institution, such as a legislature, can easily outperform a smaller and less diverse group of ultra-competent experts, such as a judicial system capped by a multimember appellate court. I hope not to press this thesis of the epistemic superiority of legislatures to a tendentious extreme. The ultimate benchmark and my ultimate aim is just to get things right—to state impartially the conditions under which, and the domains in which, the limits of reason support the allocation of lawmaking authority to one or another institution. If epistemic legalism is chronically skewed in favor of judicial lawmaking, the right response is not to introduce a skew in the other direction. Yet I believe it to be true, and the burden of the following chapters is to show, that when the conditions are stated impartially, the epistemically desirable allocation of lawmaking authority will be far less court-centric than epistemic legalism would have it.
Cumulative Epistemic Legalism? Although epistemic legalism is not usually put in these terms, one might pitch it as a thesis about the cumulative epistemic qualities of legislatures and courts—the latter proceeding in a common-law mode. On this version, even if legislatures have epistemic advantages relative to courts, the question is whether adding courts to the mix improves the overall epistemic capacities of the system. Epistemic legalism suggests that it is most likely to do so where courts proceed in common-law mode, filling in constitutional or statutory gaps with precedent, tradition, and custom. I will suggest that this version of epistemic legalism fails on two scores. First, it cannot account for the most critical cases, in which courts invalidate legislative action. Second, adding judges to the lawmaking system will make things worse, not better, under identifiable conditions. The first problem is that common-law constitutionalism is tested most severely when judges draw upon precedent or tradition or custom to overturn the enactments of legislatures, perhaps a quite recent enactment. Politically, this is the sort of case that attracts the most attention and most sharply tests the legitimacy of judicial review itself. Epistemically, this is the most dubious case for common-law constitutionalism, because legislators’ information is more recent than that of the judges who decided the precedents on which the invalidation is based. In any event, to whatever
T L R L T
13
extent those precedents do contain useful information, the legislators enacting statutes today can take the information into account; the precedents are as accessible to them as to the judges. The second problem is that, as Chapter 1 discusses, more heads can actually be worse than fewer; adding imperfect epistemic agents to the system might not merely produce diminishing returns, it might actually reduce the system’s overall epistemic quality. Adding judicial review of constitutional questions can induce a kind of epistemic free-riding on the part of legislators who anticipate that the judges will catch their mistakes, and who thus make more mistakes. Furthermore, even in statutory and regulatory domains, judges of low capacity may act as a kind of epistemic bottleneck or chokepoint, whose attempts to understand and implement the commands of epistemically superior legislatures will go badly awry.
Common-Law Constitutionalism and Its Antonyms As we have already begun to see, the major testing ground for epistemic legalism is the constitutional common law; here the stakes are highest. If the limits of reason support judicial lawmaking even in constitutional law, and even as against the policies adopted by current legislatures and executive institutions, then epistemic legalism is decisively confirmed. Thus the main line of the argument focuses on constitutional common law, insofar as it is justified on epistemic grounds. There are several antonyms to common-law constitutionalism, and I consider them all from an epistemic perspective. One antonym is originalism, or the idea that the main source of constitutional law should be the original public meaning of the constitutional text. Here the commonlaw constitutionalist emphasizes that a practice of following precedent results in “living constitutionalism,” as opposed to static enforcement of the original meaning. If the point of living constitutionalism is an epistemic comparison between precedent, on the one hand, and the framers’ knowledge, on the other, then I fully agree that living constitutionalism is superior on average. Judges acting in changing conditions are indeed more likely to track the truth by following and modifying the views of other recent judges than by following the views set forth by constitutional framers in the wildly dissimilar circumstances of the remote past. However, nothing in this implies that judges should invoke their own precedent to invalidate recent legislation on constitutional grounds. If anything, the same reasoning implies that they should not do so. Chapter 2 explores these issues in depth.
14
L L R
In other cases, common-law constitutionalism is opposed to ambitious moral and political theorizing by individual judges. Here the common-law constitutionalist emphasizes that because many judicial heads are better than one, judges are likely to do better by following precedent and tradition than they will do on their own. The limits of reason suggest that any individual judge should be cautious about imposing her pet theories, where there is contrary precedent or tradition. In this modest sense, I believe that common-law constitutionalism is plausible and, under certain conditions, even correct. Chapter 1 attempts to state those conditions, which are, however, narrower and more fragile than common-law constitutionalists often assume. In the most consequential version, however, common-law constitutionalism is opposed, not to the knowledge of the framers or to the individual judge’s theorizing, but to the conclusions of current legislatures. Here the common-law constitutionalist argues that a line of constitutional precedent under vague or ambiguous constitutional texts should trump enacted statutes, on epistemic grounds, where the two conflict. I argue that this version of common-law constitutionalism is both methodologically flawed and substantively incorrect. That many judicial heads are better than one does not at all demonstrate that many judicial heads are better than many legislative ones. And it turns out, I will claim, that legislatures enjoy comparative epistemic advantages over courts even, or especially, in matters constitutional. Overall, then, I argue for a regime of judicial deference to legislative development of constitutional law through statutes—a regime of commonlaw constitutionalism by legislatures. It is familiar that statutes can “liquidate” written constitutions by filling gaps and interpreting ambiguities, by giving content to vague or aspirational constitutional provisions, and by updating constitutional norms over time, adapting them to changing circumstances. This is the focus of chapters 2 and 3, which suggest that the very epistemic arguments typically adduced in favor of common-law constitutionalism, by self-described Burkeans and others, instead show the comparative superiority of statutes as instruments for tracking the truth about constitutional facts and values, especially under conditions of rapid social and political change. Another major alternative to common-law constitutionalism is formal constitutional amendment; as compared to either statutes or judicial decisions, amendments are harder to enact but more durable when enacted. American legal theory dismisses constitutional amendments as too difficult to obtain under the processes of Article V, ignoring that the
T L R L T
15
rate of constitutional amendment has fallen to zero in the last forty years or so (ignoring the oddity that is the Twenty-seventh Amendment, ratified two centuries after its proposal). The major change has been, not structural, but intellectual and conceptual: the spread of an epistemic argument holding that amendments risk bad unintended consequences. On this view, attempts to use the constitutional amendment process as a real alternative to common-law constitutionalism are either futile or undesirable or both. To the contrary, I argue in Chapter 5 that amendments are not futile and are no more epistemically suspect than is the constitutional common law. Indeed, the products of the amendment process are epistemically superior to those of common-law constitutionalism, under identifiable conditions.
The Ordinary Common Law and Other Sources of Law Because there are multiple strands within epistemic legalism, the word “law” in this book’s title is also eclectic. Common-law constitutionalism is only the center of the canvas, not the whole picture. Throughout, I criticize the arguments adduced by Burke, Hayek and others in favor of the epistemic superiority of tradition, custom, and the ordinary common law. Those arguments are logically flawed, rest on fragile mechanisms and peculiar assumptions, or cannot be squared with the evidence. My basic suggestion is cumulative and pitched in the alternative: epistemic legalism is suspect in the case of the ordinary common law and an outright failure in the very different case of the constitutional common law. Even if the general critique of the common law fails, the critique of common-law constitutionalism can still hold. It is possible to accept the basic thrust of Burkean and Hayekian praise for the ordinary common law without subscribing to the modern extension of those claims to the constitutional common law. I shall argue that common-law constitutionalism is, for institutional reasons, a very different enterprise than the ordinary common law, with which it shares only a name and some superficial resemblances. In the administrative state, much of law is neither common nor constitutional, but statutory; executive officials and administrative agencies charged with interpreting statutes are central actors. The executive is himself both a formal and informal participant in the legislative process, and the agencies bear frontline responsibility for filling in and carrying out broad legislative commands. Although for shorthand I will refer to “legislatures” unless there are special reasons to do otherwise, I always
16
L L R
mean to refer to legislative action in an extended sense that includes legislative decisions to enact statutes delegating lawmaking power to the executive and the administrative agencies. One of the main points of delegation, of course, is to take advantage of the epistemic specialization or expertise of the executive and the agencies. By lumping in delegated lawmaking with ordinary lawmaking, then, I am understating the epistemic capacities of nonjudicial lawmaking, thereby skewing the assumptions against the thesis I will defend. Where appropriate, I focus specifically on the independent epistemic competence of the executive and the administrative agencies, but that is not my main concern. My focus is on the relative epistemic capacities of courts, on the one hand, and of the joint legislative-executive lawmaking process, on the other. If, as I suggest, the correct epistemic account would transfer a great deal of lawmaking authority away from courts, it is a separate and further question how precisely that power should, on epistemic grounds, be allocated between and among legislatures, the executive, and the bureaucracy. Any answer to that question would be consistent with my major claims. I do not directly address statutory interpretation except in Chapter 4, which suggests that the Supreme Court should have one or more justices who are not lawyers (or at least not only lawyers), and who thus would bring their nonlegal expertise to bear on statutory as well as constitutional cases. Given the distinct importance of statutory interpretation in the Court’s work, many of the points that follow transpose comfortably to that setting, mutatis mutandis. Much of statutory interpretation proceeds in common-law–like fashion, especially for old statutes whose text is buried under a pile of precedents accumulated over decades or centuries. Where this sort of common-law statutory interpretation is justified on epistemic grounds, parallel to the epistemic justifications for common-law constitutionalism, then my negative and positive theses also hold in parallel, implying a restricted role for judges in statutory interpretation.
Bentham, Thayer, and Legal Theory Having stated my main theses and themes, I will finish by sketching some of the project’s larger background in legal and political theory. In my rogues’ gallery hang Burke, Hayek, and their modern exegetes; but portraits of Jeremy Bentham and James Bradley Thayer hang in places of honor. What unites Bentham and Thayer is skepticism about a remarkably persistent meme in legal theory: legislatures are the seat of social “will,” while
T L R L T
17
courts are the seat of epistemically impressive “judgment,” as Alexander Hamilton implied in Federalist 78.8 On this view, legislatures aggregate preferences, while courts aggregate beliefs. By contrast, Bentham and Thayer see that legislators’ capacity for judgment is formidable and is indeed, under robust and identifiable conditions, superior to that of courts. Neither Bentham nor Thayer is willing to concede the one-sided institutional claims of epistemic legalism. More broadly, one of my major claims is that updating the Benthamite critique of the common law, and transposing it to the constitutional setting, leads to Thayerian conclusions.
Bentham: Philosophical and Institutional Commitments As between the two, Bentham is by far the greater figure and must have pride of place. I begin with a distinction between Bentham’s philosophical project and his institutional project. In legal theory today, Bentham is usually mentioned only in connection with his philosophical commitments: a nascent rational-choice picture of decisionmaking; normative consequentialism; a hedonic-utilitarian value theory; and a purely aggregative social welfare function, summing pleasures and pains across persons. Bentham thus serves as a foil for any philosopher or legal theorist who wants to propose a more complex account in any of these domains. Some theorists go so far as to dismiss Bentham as merely an unsystematic precursor of modern law-and-economics—the research program, starting in its modern form in about 1960, that applies economic tools to legal theory. The connection is that economics is a rationalist discipline and, in its normative dimension, a consequentialist and welfarist one. When Judge Richard Posner attempted to sum up Bentham’s influence on the “law and economics” movement, he focused on utilitarianism rather than on Bentham’s views about the institutional allocation of lawmaking authority, which he barely mentioned.9 I think it is wrong to see Bentham in this light. Bentham’s main commitments were institutional, not philosophical or economic; his claims about such matters were byproducts of his lifelong campaign for codification and legal reform. Bentham’s main concern was not to offer positive hypotheses about the effects of legal rules, or even normative claims about how to design those rules so as to maximize utility,
8 9
See The Federalist No. 78 (Hamilton) 437–40 (Isaac Kramnick ed., 1987). Richard A. Posner, Bentham’s Influence on the Law and Economics Movement, 51 Current Legal Probs. 425 (1998).
18
L L R
although he also did those things. Rather his project was to rationalize and demystify English law by exposing the obscurantism and social disutility of the common law.10 Many, although by no means all, law-and-economics scholars celebrate the common law and argue for its efficiency, either on the ground that common-law judges intentionally maximize efficiency or on the ground that invisible-hand mechanisms of the litigation process weed out inefficient precedents. But institutional views that favor or for that matter oppose the common law, as against other suppliers of lawmaking and processes of lawmaking, have no necessary connection to the methods of law and economics and cannot be derived from those methods without supplemental premises. In contrast to law-and-economics scholars who celebrate the common law, I believe that Bentham’s institutional commitments are at least as important as his methodological and philosophical ones and deserve a fresh hearing both in modern law-and-economics and in modern legal theory more generally. None of this implies that Bentham’s institutional critique of the English common law was solely or even centrally based on epistemic grounds. Bentham objected to many features of the common law and the professional sociology of the English bar, including the excessive complexity and radical uncertainty that a common-law system generated, the jargon-laden and antiquarian character of the common law, and the pompous charade of natural law and immemorial custom masking lawyerly self-interest. As the quotation at the outset indicates, however, another important strand in Bentham’s critique was an argument for the superior epistemic capacities of legislatures. This strand is persistent, although not central, in Bentham’s critique of Sir William Blackstone, but became central when Bentham later turned to the design of legislatures and lawmaking institutions.
Updating Bentham Methodologically, then, I aim to modernize the institutional side of the Benthamite project. Because of the widespread support, express or tacit, for epistemic legalism among modern Anglo-American legal theorists, Bentham’s critique of the common law has failed to keep up with the progress of institutional knowledge; it requires updating along two margins. 10
H.L.A. Hart, Bentham and the Demystification of the Law, 36 Mod. L. Rev. 2 (1978).
T L R L T
19
First, the tools of modern social science, especially political science, psychology and economics rather than political philosophy, must be brought to bear on Bentham’s institutional commitments and conclusions, to help us determine which are valid and which are not, and under what conditions. Second, the Benthamite critique must be translated into a system based on “constitutional common law”: the view that constitutional law is, and should be, mostly a judge-made product, with precedent trumping the constitutional views of legislatures and executive officials if necessary. Of course there was no real equivalent to the constitutional common law in Bentham’s world; despite strands of natural-law constitutionalism in early cases and in Blackstone, the ultimate postulate of the English legal system was that judges were bound to obey an express command of Parliament. One must therefore extrapolate and update the institutional side of Bentham to see if the critique of common law holds in the constitutional setting. I will argue that it does hold, in spades. The most striking rationale for common-law constitutionalism is epistemic; and that rationale fails, as I shall argue throughout the succeeding chapters. Indeed, whatever the epistemic credentials of the “ordinary” common law, the constitutional common law fares less well in this regard, precisely because it relies upon the epistemic abilities of a relatively small number of judges, with a distinctive and somewhat parochial informational base, to trump the products of large and diverse modern legislatures. This epistemic updating of Benthamism into a critique of the constitutional common law is the centerpiece of my project.
Condorcet and Bentham vs. Burke and Hayek Part of the project of epistemic legalism has been to appropriate, for the defense of common law and common-law constitutionalism, the thinking of the Marquis de Condorcet, particularly his famous Jury Theorem.11 I explain the Jury Theorem in detail in Chapter 1; roughly, it shows that where there is a binary choice as to which a right answer exists, a majority vote among a group of sincere voters who are better than random will approach perfect accuracy as the size of the group increases, as individual accuracy increases, and as the correlation of biases within the
11
Marquis de Condorcet, An Essay on the Application of Mathematics to the Theory of Decision Making (1785), reprinted in Condorcet: Selected Writings 33 (K. Baker ed., Bobbs-Merrill 1976).
20
L L R
group decreases. Chapter 2 explains the Condorcetian interpretation of Burke, and of common-law constitutionalism, showing that common-law constitutionalists have explicitly or implicitly attempted to shore up epistemic Burkeanism by invoking Condorcetian logic.12 Throughout, I reject that enterprise. Although the details are laid out in succeeding chapters, my main point is that the Condorcetian argument for common-law constitutionalism gets the Jury Theorem’s institutional implications backwards. A major implication of the Jury Theorem is that it undermines the epistemic credentials of experts as compared to sufficiently large groups of nonexperts, where the views of the nonexperts are at least better than random. The “wisdom of crowds”13 can, under plausible conditions, prove superior to the wisdom of experts. But this includes legal experts and legal expertise as well as other forms of expertise. Even where legal right answers exist, a large decisionmaking group that is better than random can outperform a group composed solely of lawyers. Moreover, I argue in Chapter 4 that adding nonlawyers to a multimember court can actually produce a greater chance of obtaining correct legal answers, either in cases where law itself incorporates nonlegal knowledge, or in cases where the right answer is in some sense “internal” to law. The main engine that drives this surprising result is that lawyers have correlated biases, arising either from their professional training or from the self-selection of certain types into the legal profession. As I explain in Chapter 4, reducing the correlation of biases by introducing decisionmakers from other professional backgrounds can more than compensate for the loss of purely legal expertise. From this perspective, Condorcet is a more natural ally of Bentham14 than of common-law theorists like Burke and Hayek. Condorcetian themes connect better with Benthamite suspicion of legal claims to expertise than with the claims of common-law constitutionalists that a tradition of judicial decisionmaking embodies collective wisdom or implicit knowledge. As the quotation set out at the beginning suggests, a Benthamite perspective is alert to the possibility that a large group of legislators can do better than judges at discerning “common reason.” Here we must be careful; it is ambiguous whether the right answer that Bentham is discussing—the content of “common reason”—is right in the sense that it exists independently of anyone’s preferences, or whether 12 13 14
See, e.g., Cass R. Sunstein, Burkean Minimalism, 105 Mich. L. Rev. 353 (2006). James Surowiecki, The Wisdom of Crowds (2004). For the possibility that Bentham read and was influenced by Condorcet, see Jon Elster, The Optimal Design of a Constituent Assembly (draft on file with author).
T L R L T
21
instead Bentham is just saying that legislators are more likely than judges to be correct about what a majority of people (or voters) would prefer. As we will see in Chapter 1, however, the latter interpretation is perfectly consistent with the Jury Theorem. The right answer required by the Jury Theorem need not be a right answer independent of majority preferences, although it can be.
Condorcet and Thayer vs. Hamilton A corollary is that I aim, throughout, to supply Condorcetian underpinnings for the Thayerian position in constitutional law, which counsels deference to legislatures in constitutional adjudication. One of my claims is that the limits of reason, rightly understood, support that position. Once common-law constitutionalism is justified on epistemic grounds, as theorists of epistemic legalism would have it, then the superior epistemic capacities of legislatures undercut any basis for overriding legislative judgments. That point is static; but Thayer also had a striking dynamic insight, much in advance of his time, that the very existence of constitutional judicial review might have pernicious effects on legislative output. Thayer’s idea was that the “judicial overhang”15—the anticipation by legislators that judges would review their work—might distort legislative behavior for the worse. Although this effect can be interpreted in nonepistemic terms, as a form of legislative buck-passing or moral hazard, I will advance an epistemic interpretation in chapters 1 and 2. Beyond these claims, Thayer’s larger importance is that in an intuitive and undertheorized fashion he appreciated the hollowness of the standard Hamiltonian picture in constitutional law—the idea that the courts embody “judgment,” while the legislative and executive branches embody “will.” In Thayer’s formulation, “judgment,” far from being the exclusive province of the courts, is precisely the distinctive virtue that legislatures bring to their task. So another part of my project here is to supply social-science underpinnings for Thayer’s basic insight into the epistemic capacities of legislatures. What I propose is a modernized and epistemic version of Thayerism.
Other Precursors Although the core of the argument stems from Bentham, Thayer, and (in a more qualified fashion) Condorcet, I also draw upon the work of a small 15
Mark Tushnet, Taking the Constitution Away From the Courts (1999).
22
L L R
band of neo-Benthamites and quasi-Benthamite fellow travelers writing in legal theory. Especially useful is the work of Jeremy Waldron, who offers an appealing picture of the “dignity of legislation” and who has done pioneering work on the epistemic credentials of legislatures; the influence of his work will be apparent throughout. Methodologically, however, Waldron pursues his arguments principally in a philosophical and jurisprudential vein. His explicit aim is to offer an idealized picture of legislatures to offset the idealized picture of courts that is common in legal theory.16 My aim is to work on these questions with pragmatic tools drawn from social science, rather than the conceptual tools of analytic and political philosophy, and to offer a realistic, rather than idealized, epistemic comparison of all relevant institutions.
“Justices and Company”: The Sociological Substructure of Law Finally, while I aim to engage the theoretical underpinnings of epistemic legalism, I do not neglect its sociological substructure. Indeed, to do so would be distinctly un-Benthamite; one of Bentham’s major preoccupations was with the politics and sociology of “Judge and Company,” the English bench and bar. Bentham was inclined to diagnose the motivations of his opponents, which led him to produce a nascent but rather crude theory of ideology in which the excesses and failures of the common law could be traced to “sinister interest” on the part of the lawyerly class. I have strictly avoided the lure of ideological critique here, and lawyerly self-interest or group interest is no part of my argument; epistemic legalism is an intellectual error, and getting things right intellectually shows the affirmative virtues of the codified Constitution. But that does not at all require us to ignore the sociological questions. Rather, we must ask how the sociological configuration of Anglo-American legal systems themselves affect the epistemic competence of various institutions and the personnel who staff them. I take up this issue in Chapter 4, which attempts to update the sociological side of Bentham’s institutional project. Just as Bentham sought to break up Judge and Company, so too I suggest that we might do well to break up Justices and Company—the professional monopoly of common-law constitutionalism by the lawyers who exclusively staff the Supreme Court. 16
Jeremy Waldron, Law and Disagreement 31–32 (1999).
T L R L T
23
Thus the basic suggestion is that it is bad, on epistemic grounds, that all of the Supreme Court’s members are lawyers. Nothing in law or elsewhere requires this, and in a world of substantial common-law–making by the justices (both in nominally constitutional and in nominally statutory cases), it is a bad idea to have all the Court’s membership drawing upon the same, highly correlated sources of professional training and information. This is a recipe for lawyerly groupthink. On epistemic grounds, sensible institutional design demands a modicum of professional diversity among the arbiters of the law. None of this requires us to impugn the motivations of lawyers or of the Court’s members. It only requires a willingness to see the epistemic benefits of professional diversity along with other types of diversity.17
17
See generally Scott Page, The Difference (2007).
This page intentionally left blank
Chapter 1
Many-Minds Arguments
I
begin with the first major strut of epistemic legalism: many-minds arguments, which have a long pedigree in a relatively untheorized or implicit form, but which have lately been made more explicit in many sectors of legal theory.18 Judicial precedents are said to aggregate the collective knowledge of many judges over time; customs and traditions incorporated into law are said to embody the collective wisdom of the broader society; the law of foreign states is said to provide aggregated information about optimal legal rules; deliberation in legislatures and direct democracy embodies the “wisdom of the multitude.”19 The theme that unites these arguments is a claim that in some way or another, many heads are better than one. But the genus of many-minds argumentation is internally heterogeneous and contains many species, including arguments about how legal and political institutions aggregate information, evolutionary analyses of those institutions, and analyses of the virtues and vices of deliberation. 18
19
The phrase “many minds” comes from Cass R. Sunstein, Infotopia (2006); see also Cass R. Sunstein, A Constitution of Many Minds (forthcoming 2009). I adopt it as a useful shorthand for the ancient idea that many heads are better than one. So dubbed by Jeremy Waldron. See Jeremy Waldron, The Dignity of Legislation 93 (1999). The arguments referred to here are documented in the body of the paper.
25
26
L L R
The chapter has three aims. First, I introduce many-minds arguments for use throughout the subsequent discussion, especially in chapters 2, 3, and 4. I provide an intellectual zoology, distinguishing among the major species of many-minds arguments and explaining the conditions under which one or another is more or less successful. Second, I provide grounds for skepticism about the robustness and generality of manyminds arguments. Such arguments are typically stylized, pitched at a high level of abstraction, rest on fragile mechanisms, and apply only under a narrow range of conditions. They are the hothouse flowers of legal theory, attractive to the eye but usually too feeble to survive in the cold wide world. Third, I document that many-minds arguments have been invoked, for the most part, in support of epistemic legalism—that is, to support the common law vis-à-vis legislation, to support the constitutional common law vis-à-vis statutes, and more generally to support judge-made law as against lawmaking and law-interpretation by legislatures and agencies. That is not their only possible use—the wisdom of the multitude, for example, is a many-minds argument that has been invoked to demonstrate the epistemic competence of legislatures—but that is itself one of the central points I will pursue in this chapter and later: many-minds arguments, properly confined and understood, give at least as much reason to favor legislative lawmaking as judicial lawmaking. Not only are they usually fragile, but where they are robust, they cut against epistemic legalism rather than in its favor.
Many Minds, Many Problems There are four quite general problems that cut across the different categories of many-minds arguments, and that make such arguments less robust and general than they appear at first glance: (1) Whose minds?: The group or population whose minds are at issue is often equivocal or ill defined. The multiplicand or denominator always matters for a many-minds argument addressing particular legal institutions; clarifying the denominator can make a plausible-seeming argument less impressive. (2) Many minds, worse minds: The quality of minds is not independent of their number; rather, number endogenously influences quality, often for the worse. More minds can be systematically worse than fewer because of selection effects, incentives for epistemic free-riding, and emotional and social influences.
M-M A
27
(3) Epistemic bottlenecks: The epistemic benefits of many minds are often diluted or eliminated because the structure of institutions funnels decisions through a small group of decisionmakers who occupy an epistemic bottleneck or chokepoint. There is unavoidable epistemic slack between many minds and their few leaders; epistemic agenda-setters are also unavoidable. In general, there is an epistemic analogue to the iron law of oligarchy. (4) Many minds vs. many minds: The insight that many heads can be better than one gets little purchase on the institutional comparisons that pervade legal theory, which are typically many-to-many comparisons rather than one-to-many. To be clear, my suggestion is not that many-minds arguments have zero utility. Rather, these problems create a succession of filters. Given an initial set of many-minds arguments, each filter will weed out some of those arguments, showing them to be either ill-specified or unpersuasive. Some subset of many-minds arguments will survive all four filters and thus prove both well-specified and plausible, and thus useful for legal theory. Indeed, in Chapter 4, I will offer a many-minds argument, rooted in the Jury Theorem and in deliberation, for appointing nonlawyer justices to the Supreme Court. I hope, and claim, that the argument of Chapter 4 is one that survives the filters identified here, but the proof will be in the pudding. Above all, there is no necessary connection between the epistemic virtues of many minds and a preference for judge-made law over law made by legislators and executive officials. Quite the contrary, under an identifiable and robust set of conditions, which I will lay out in chapters 2 and 3, many-minds arguments suggest that judicial lawmaking is comparatively unpromising on epistemic grounds, relative to lawmaking by the other institutions of the legal system. Many minds need not be, and often should not be, judicial minds.
Many-Minds Arguments: A Taxonomy Before examining these problems in detail, we need to get clear about the nature of many-minds arguments. Why, exactly, might many heads be better than one? There are three main reasons: the aggregate judgment of many might employ dispersed information better than the judgment of one; the judgments of many heads, over time, might weed out bad policies or institutions through an evolutionary process; finally, deliberation and argument among the many might contribute diverse perspectives, resulting
28
L L R
in better policies or institutions than any one could devise. I will call these the arguments from information aggregation, evolution, and deliberation, and associate them with Condorcet, Hayek and Aristotle, respectively. In many bodies of legal theory, information aggregation and evolution are intertwined with the idea that legal and political actors would do well to follow tradition, and tradition is associated in turn with Burke. In chapters 2 and 3, I suggest that traditionalism can be interpreted in either informational or evolutionary terms, and afford extended separate treatment to Burkean theory in law and politics.
Information Aggregation and Condorcet The most widely invoked model of information aggregation is the Condorcet Jury Theorem. In its simplest form, the Jury Theorem states that where there are two alternatives, one of which is correct (somehow defined; we will see that correct need not mean “correct independent of people’s preferences”) and where in a choice between those alternatives the members of a group are even slightly more likely to be right than wrong (and are thus better than random), then as the number of members in the group increases, the probability that a majority20 vote of the group is correct tends towards certainty. Two implications are that the group’s “competence” or chance of being correct can exceed that of the group’s most competent members, and that a large enough number of fairly poor (but better than random) guessers can easily prove more competent than a small panel of highly competent experts. Hence recent suggestions that the Jury Theorem models “the wisdom of crowds”21 or “smart mobs.”22 A corollary 20
21 22
The Jury Theorem can be extended to qualified majority (“supermajority”) rules, but only with restrictions. It has been shown that qualified majority rules maximize the probability of making a correct decision, but only if the status quo is stipulated to prevail in the event that no alternative garners the requisite supermajority. See Ben-Yashar and Nitzan, The Optimal Decision Rule for Fixed-Size Committees in Dichotomous Choice Situations: The General Result, 38 Int. Econ. Rev. 175 (1997). This is of course a democratically suspect condition. If the status quo preference is abandoned, then a weaker result holds: “for sufficiently large electorates. . . if the average competence of the voters is greater than the fraction of the votes needed for passage, a group decision is more likely to be correct than the decision of a single randomly chosen individual.” Mark Fey, A Note on the Condorcet Jury Theorem with Supermajority Voting Rules, 20 Social Choice and Welfare 27 (2003). This condition is rather demanding; if the decisionmaking group uses a 2/3 majority rule, for example, then average competence must be at least. 67. Nothing in my claims here turns on whether the group voting rule is a simple or qualified majority, so I will refer to “majority” voting for simplicity. James Surowiecki, The Wisdom of Crowds (2004). Howard Rheingold, Smart Mobs (2002).
M-M A
29
is that diversity—here, the statistical independence of the guesses—makes a big difference in group performance, holding competence constant.23 Moreover, the Jury Theorem can be extended in several directions without serious loss of generality. In the simplest version the assumption is that all guessers have the same competence, but it goes through in much the same way if the group is heterogeneous but the average competence is better than random.24 Where that is so, a majority of the group will, given the other conditions, prove more competent than the average individual and perhaps even more competent than the most competent individual. The Jury Theorem has also been extended to more than two options, in which case the option that achieves plurality support is more likely to be correct than any other option, though not necessarily more likely than all the incorrect options combined.25 All this has provoked large claims about the Jury Theorem’s significance. I will briefly mention two serious and largely unresolved conceptual problems with the Theorem, and then comment on its interpretation. Conceptually, there are at least two respects in which the Theorem is still poorly understood. First, a crucial engine behind the Jury Theorem is the independence of the group members’ views or guesses. Even if particular voters make biased guesses, the group as a whole will be unbiased on average if guessers’ biases are uncorrelated, which is the nub of independence. In one interpretation, this washing out of uncorrelated biases is what Rousseau was getting at when he claimed that the general will or common interest emerges when and because the differences between individual wills cancel out.26 The issue of correlation is just as important as the issue of accuracy;27 indeed, introducing voters who are worse than random can
23 24
25 26
27
Scott Page, The Difference (2007). And so long as the distribution of competence is symmetric around the mean. See Bernard Grofman et al., Thirteen Theorems in Search of the Truth, 15 Theory & Decision 261, 273–4 (1983). This condition implies that the Theorem does not work in some fairly common cases. Take a group of three voters whose competences are .26, .26. and 1. The mean competence is greater than .5, but the asymmetric distribution is fatal. The first two voters will join forces to override the third more than half the time (.74 x .74 = .54), even though the third voter is, by stipulation, always correct. Christian List and Robert E. Goodin, Epistemic Democracy: Generalizing the Condorcet Jury Theorem, 9 J. Pol. Phil. 277, 286 (2001). See Bernard Grofman and Scott L. Feld, Rousseau’s General Will: A Condorcetian Perspective, 82 Am. Pol. Sci. Rev. 567 (1988); for further discussion, see the symposium on Grofman and Feld’s paper in APSR, David M. Estlund, Jeremy Waldron, et al., Democratic Theory and the Public Interest: Condorcet and Rousseau Revisited, 83 Am. Pol. Sci. Rev. 1317 (1989). This is the main point of Page, supra note 23.
30
L L R
improve the group’s overall performance if the new voters sufficiently reduce correlation across the group as a whole.28 What is unclear is whether, and to what extent, independence is compromised by common deliberation or discussion. Rousseau feared that it was, and therefore recommended voting without public deliberation; others deny that deliberation has such an effect.29 To the extent that deliberation compromises independence, the epistemic power of groups will be reduced, although not eliminated; nonindependent guesses simply count as zero, so independence is a matter of degree, not an on-off switch. The independence required by the Jury Theorem is “statistical, not causal,” meaning that so long as A’s vote is the same as A’s vote conditional on B’s vote, statistical independence is preserved and the Jury Theorem goes through.30 This is a useful clarification but does not solve or even purport to solve the important problem, which is whether and when causal dependence—such as the guessers having a common social background, or common training, or deferring to a common opinion leader— undermines statistical independence. Absent any general account of this, the basic reach of the Jury Theorem is not well understood31 and no amount of possibility theorems or anecdotes about wise crowds will tell us whether the Jury Theorem is an important tool of political and legal theory or a minor curiosity. Empirically, the social psychology literature is on the whole pessimistic about the wisdom of crowds, finding that “groups excel as judges only under limited conditions.”32 28
29 30 31
32
Krishna K. Ladha, The Condorcet Jury Theorem, Free Speech, and Correlated Votes, 36 Am. J. Pol. Sci. 617, 628–29 (1992); Krishna K. Ladha and Gary Miller, Political Discourse, Factions, and the General Will: Correlated Voting and Condorcet’s Jury Theorem, in Collective Decision-Making 393, 403–05 (Norman Schofield ed., 1996). See Estlund, Waldron, et al., supra note 26, at 1326–27. David M. Estlund, Democratic Authority: A Philosophical Framework 225 (2007). See id. at 226. For some comparative statics of the relationship between competence and deference to opinion leaders, see David Estlund, Opinion Leaders, Independence, and Condorcet’s Jury Theorem, 36 Theory & Decision 131 (1994). Daniel Gigone and Reid Hastie, Proper Analysis of the Accuracy of Group Judgments, Psychological Bulletin 121 (1997) 149–67, at 149. Reviewing a large number of studies, the authors summarize the results as follows: For the most part, group judgments tend to be more accurate than the judgments of typical [i.e. median] individuals, approximately equal in accuracy to the mean judgments of their members, and less accurate than the judgments of their most accurate member. . . . The judgment tasks in which groups can consistently outperform individual judges may therefore be extremely limited. Id. at 153. The nature of the required judgment makes a difference; groups performing “eureka” tasks, whose solution is demonstrably correct once stated, tend to approach the performance of their best members, whereas groups performing other sorts of tasks “tend to perform at the level of their average members.” Id. at 149.
M-M A
31
Furthermore, there is a conceptual puzzle about what it means to say, in jury-theoretic analyses, that the guessers are better than random. The problem is that the baseline against which randomness is measured is affected by how the options are partitioned.33 (There is a related problem of partition of states that afflicts the principle of insufficient reason in the theory of decisionmaking under uncertainty).34 If there might be 100, 200, or 300 beans in the jar, should we say that there are three choices, so that for a random guesser there is a 33 percent of each being right? On the other hand, we could say that the choices are “100 or anything else”; would this mean that for a random guesser there is a 50 percent chance that 100 is correct? And of course we could do the same for 200 or 300. This problem suggests that application of the Jury Theorem, in multiple-option cases, might be sensitive to the order in which choices are presented, creating the same problems of agenda-setting and path dependence for judgment aggregation that are present, by virtue of Arrow’s Theorem, in preference aggregation. I touch upon some problems of epistemic agenda-setting below. From the intrinsic difficulties of the Jury Theorem, I now turn to its interpretation. The Theorem is in many respects much thinner than some incautious discussions recognize. It is a strictly mathematical artifact that does not necessarily say anything about information, or the aggregation of common judgments, or the wisdom of crowds, although in certain circumstances it can be interpreted to do so. The Jury Theorem is consistent with several very different probability models, and these models yield interpretations of the Theorem that differ materially.35 I will highlight two common misconceptions. First, it is not the case that the Jury Theorem presupposes the existence of an exogenous “right answer,” where by exogenous I will mean independent of the preferences held by the group’s members (or the members of some larger underlying group). One interpretation of the Jury Theorem is the “polling model”:36 as the sample drawn out of some larger population increases, and the other assumptions of the Theorem are met, it is increasingly likely that a majority vote of the sample group tracks what a majority of the underlying population would choose. Here, if we like, we can say that the right answer is what the whole population would 33 34 35 36
See Estlund, supra note 30, at 229–30. See references in Adrian Vermeule, Judging Under Uncertainty 173–75 (2006). Paul H. Edelman, On Legal Interpretations of the Condorcet Jury Theorem, 31 J. Legal Stud. 327 (2002). Id. at 332–33.
32
L L R
choose, and there is an attenuated sense in which the members of the polled subgroup have “dispersed information” about that, but that is not the sense of right answers or dispersed information that provokes interest in the Jury Theorem. Second, and relatedly, the Jury Theorem does not necessarily aggregate information at all; it has no intrinsically epistemic properties. In one illuminating model there is no dispersed information being aggregated through a group attempt to answer a common question with a right answer independent of preferences, such as the number of beans in a jar. Rather the group members are each, severally, asking which collective choice will best promote the satisfaction of their own preferences.37 In this model, as the number in the group increases, it becomes near certain that the majority is correct about which collective choice will maximize satisfaction of the preferences of the majority. People sometimes talk as though there is a necessary connection between the Jury Theorem and the aggregation of collective judgments about a common preference-independent question, as opposed to the aggregation of individual preferences. There is no such necessary connection, as the polling model and the preference-aggregation model show. The upshot is that it is wrong to say that the Jury Theorem “requires” that voters be asking a common question, in any substantial sense. In the preference-aggregation model, there is a trivial distributive sense in which group members are asking the same question—each is asking: “What will maximize the satisfaction of my preferences?”—but again this is not the collective sense of a common preference-independent question that inspires interest in the Jury Theorem among epistemic democrats. What is true is that insofar as we are interested in the aggregation of judgments, as opposed to preferences, the group members must be asking the same question; if they are asking different questions, the Jury Theorem is not the right model of aggregation, although other models may be available (as in the Aristotelian models discussed shortly). However, nothing in the logic of the Jury Theorem is inherently tied to the aggregation of judgments, as opposed to preferences. All the Jury Theorem does is to apply the Law of Large Numbers to a population of statistically independent guesses. These may be guesses informed by dispersed information, or they may be statements of preference, or what-have-you. The logic of the Jury Theorem is unaffected by the
37
Nicholas R. Miller, Information, Electorates, and Democracy: Some Extensions and Interpretations of the Condorcet Jury Theorem, in Information Pooling and Group Decision Making 173 (Bernard Grofman and Guillermo Owen eds., 1983).
M-M A
33
particular interpretation. It is true that in particular cases the Jury Theorem can be applied to model the aggregation of information, but it need not have anything to do with that. At a minimum, the analyst must be careful to specify what exactly is being aggregated and must be careful not to equivocate between different interpretations.38
Evolution and Hayek In legal theory, evolutionary many-minds arguments are likely to focus either on Hayek or Burke, whose ideas on this subject overlap to some degree. I will introduce Hayek here; Chapter 2 offers an overview and critique of legal Burkeanism, while Chapter 3 offers a fuller account of evolutionary arguments in Burkean and Hayekian legal theory. For my purposes, we may associate Hayek with two major ideas: the division of knowledge39 (the epistemic analogue to Adam Smith’s division of labor) and an evolutionary argument for the virtues of the common law.40 The first emphasizes the dispersed and tacit character of knowledge in markets, an argument that Hayek famously used to win the debate over the feasibility of socialist calculation—whether a socialist planner could acquire and process sufficient information to make efficient centralized decisions. The second celebrates the common law by an extended analogy to markets, an analogy that is shaky at best. Hayek’s central claim about explicit markets was true, important, and largely original, which led him to draw ever-more-extended analogies between markets and other social institutions, such as custom, morality, and common law. The analogies are evocative, which accounts for his enduring appeal. But Hayek was an unsystematic thinker who had a strong tendency to think that all things he counted as good—spontaneous order, the division of knowledge, markets, morality, custom, and the common law— somehow all go together. They do not; Hayek’s analogies are flawed, or so I will briefly suggest. Moreover, the version of his thought about the common law on which he seems to have finally settled—that the common law was socially beneficial as a system in competition with other systems, rather than efficient taken rule-by-rule—strips his defense of the common law of any implications for action within an ongoing common-law system. 38 39 40
Edelman, supra note 35, at 339–48, provides several examples of arguments that equivocate between different interpretations of the Jury Theorem. See Friedrich A. Hayek, The Use of Knowledge in Society, in Individualism and Economic Order 77–91 (1945). Friedrich A. Hayek, Law, Legislation, and Liberty, Vol. I: Rules and Order (1973).
34
L L R
We must start with some distinctions to come to grips with Hayek. One distinction is between intentional mechanisms and invisible-hand mechanisms. In the Jury Theorem model of information aggregation, participants are trying to get a particular answer (whether the preferenceindependent right answer or the answer that would maximize the satisfaction of their own preferences). The epistemic value of their collective effort—the many-minds argument—is the intended result of their actions, although the group competence may of course differ from individual competence through the miracle of aggregation. In evolutionary manyminds arguments, by contrast, the epistemic value of many minds arises as a byproduct of actions taken for other reasons. Evolutionary many-minds arguments are thus a species of invisible-hand arguments,41 in which the epistemic competence of the group is “the result of human action, but not of human design.”42 Hayek’s core commitment is to spontaneous order, which means to invisible-hand mechanisms (although as we will see Hayek seems to have assumed that all invisible-hand mechanisms must be evolutionary, which is not the case). The Condorcetian mechanism is a model of aggregated intentions, not an invisible-hand mechanism, whereas Hayek thinks that the aggregation of information must occur through the action of the invisible hand. In this sense, there can be no Condorcetian interpretation of Hayek. Within the category of invisible-hand mechanisms, there is a further distinction between aggregative and evolutionary mechanisms.43 Aggregative mechanisms are synchronic, explaining the emergence of an equilibrium at a given time; markets are an example. Evolutionary mechanisms are diachronic, explaining not the emergence but rather the maintenance and development of order over time. Thus, in biology, Darwin’s mechanism of natural selection requires variation, heritability, and differential reproductive fitness of the genotype (acting through the phenotype), but natural selection is entirely agnostic about the origin of the genotype, which may arise through random mutation. The point is to explain the survival, or not, of the genotype over time. Hayek seems to have systematically conflated these two types of invisible-hand mechanisms. Starting from an epistemic version of Adam Smith, inspired by aggregative markets, he slid over into an evolutionary 41 42 43
Edna Ullmann-Margalit, The Invisible Hand and the Cunning of Reason, 64 Soc. Res. 181, 186 (1997). Paraphrasing Adam Ferguson. Adam Ferguson, An Essay on the History of Civil Society 119 (Fania Oz-Salzberger ed., Cambridge Univ. Press 1996) (1767). The next few paragraphs draw heavily upon Ullmann-Margalit, supra note 41.
M-M A
35
account of morality, custom, and the common law without realizing that the subject had changed. Intrinsically, the price-system strand in Hayek’s thinking has nothing to do with evolution. When Hayekians emphasize the dispersed character of information, and the ability of market-like mechanisms using explicit prices (such as so-called “prediction markets”) to aggregate that information, their arguments are synchronic rather than diachronic, and thus not evolutionary at all. Separately, Hayek and the Hayekians have a commitment to evolutionary accounts of morality, custom and law but this commitment need not follow from the commitment to aggregative spontaneous orders. Spontaneous orders arising at a given time and extended over space, such as a market, are analytically different than spontaneous orders evolving over time, and there is no simple analogy between the two. To be sure, Hayek also means to offer an evolutionary account of markets as adjusting efficiently, over time, to exogenous shocks arising from technical, economic, and social change, and such an account is fully in the spirit of Austrian economics. But an account of that sort needs separate justification and cannot be derived from the basic Hayekian insight that the price system aggregates dispersed information, which is a strictly synchronic insight taken by itself. Below, I will examine an institutional analogue of these problems: the distinction, crucial for many-minds arguments in legal theory, between simultaneous and sequential decisionmaking. Finally, of course, spontaneous orders of either variety may either be desirable or not.44 Hayekians acknowledge this in their cautious moments but seem to hold at least a presumption that spontaneous order is functionally beneficial. But there is no such presumption. In the case of aggregative invisible-hand mechanisms, the only requisite is that a patterned social structure arises as a result of human action but not through human design. Collectively suboptimal patterns fit this requirement just as well as collectively optimal ones. The Tragedy of the Commons is a spontaneous order, just a bad spontaneous order that produces a structured social pattern of depletion and waste. What about evolutionary invisible-hand mechanisms? Here there is a strong temptation to invoke “the test of time” and to think that there is at least a presumption that evolved patterns must be functionally beneficial in some sense or another. However, within the domain of evolutionary invisible-hand mechanisms, biological evolution is very different than 44
Robert Sugden, Spontaneous Order, in The New Palgrave Dictionary of Economics and the Law 485, 494 (Peter Newman ed., 1998).
36
L L R
social or economic or cultural evolution. Here too Hayek seems to have overlooked a crucial distinction. The rate of change in the biological (“natural”) environment is, plausibly, much slower than in the social, economic, or technological environment. The slow pace of environmental change in biology makes it plausible to think that natural selection produces organisms well adapted to their ecological niches. If the rate of change in social, economic, and technological environments is high, however, then social evolution faces a shifting target: even if social structures constantly evolve towards efficiency, they may at any particular point remain very far from it.45 Once again the diachronic/synchronic distinction is crucial: the trend towards efficiency over time just does not support the very different assumption or presumption that observed institutions are beneficial at a given time. I expand upon these issues, with particular reference to Hayek’s evolutionary account of the common law, in Chapter 3.
Deliberation and Aristotle A different many-minds model altogether arises from Aristotle’s Doctrine of the Wisdom of the Multitude: For the many, of whom each individual is not a good man, when they meet together may be better than the few good, if regarded not individually but collectively, just as a feast to which many contribute is better than a dinner provided out of a single purse. For each individual among the many has a share of excellence and practical wisdom, and when they meet together, just as they become in a manner one man, who has many feet, and hands, and senses, so too with regard to their character and thought. Hence the many are better judges than a single man of music and poetry; for some understand one part, and some another, and among them they understand the whole.46 This is usually given a deliberative interpretation; the idea is that partial perspectives are shared by talking together. But deliberation is not inherent in and is not necessary to Aristotle’s idea. I mean this as a conceptual point, not an exegetical one. (Exegetically, the passage above 45 46
See Jon Elster, Ulysses and the Sirens 6 (rev. ed. 1984). Aristotle, Politics bk. 3, ch. 11, reprinted in The Politics 66 (Stephen Everson ed., 1988).
M-M A
37
says nothing explicit about deliberation, as opposed to simply “meeting together,” which is equally consistent with noncommunicative aggregation. Rather, the usual deliberative reading derives from the larger context of Aristotle’s views.47) Just as the Jury Theorem is a statistical mechanism that can go through whether or not the participants talk to one another (although their talking to one another can cause the Theorem’s conditions to hold or not to hold), so too the pooling of perspectives by itself need have nothing at all to do with deliberation or communication. In simple models of information aggregation without communication, the pooling of perspectives can be accomplished just by having each member of the group simultaneously state her partial perspective and using some aggregation rule to combine them into a whole picture.48 Despite the absence of communication, this is still not a Condorcetian model; by hypothesis no one in the group can possibly be correct about the whole picture, so there is no analogue to the Condorcetian issue of individual competence. We need to distinguish among (1) Condorcetian models of information aggregation, (2) pooled-perspective models of information aggregation, and (3) deliberation. There are no necessary connections among any of these. Deliberative interpretations of Aristotle correctly point out the differences between (1) and (2) and between (1) and (3), but elide the difference between (2) and (3).49 However, let us put this problem aside and focus on a deliberative interpretation of Aristotle, in which the pooling of perspectives is effected by communication. We are beginning to understand the conditions under which, and mechanisms by which, deliberation may produce either error or extremism.50 Those two ways in which deliberation can go bad are not conceptually the same, although they may be empirically connected. If in many environments the extreme policy is also an erroneous policy, somehow defined, then extremism is causally connected to error. This is itself an Aristotelian theme—under uncertainty, we may often do well to adopt the Golden Mean. Apart from extremism and error, however, there is a third important way in which group deliberation can go bad: deliberation may produce simple incoherence—a mishmash of views or a misshapen policy or 47 48 49 50
Thanks to Adriaan Lanni for helpful tutoring on this question. Page, supra note 23, at 185–88. See Estlund, supra note 30, at 231–32. See, e.g., Cass R. Sunstein, Deliberative Trouble? Why Groups Go To Extremes, 110 Yale L.J. 71 (2000).
38
L L R
an interminable debate that never decides anything. In any of these cases, the results of deliberation might not even rise to the level of being wrong, let alone right, however error is defined. To be either extremist or erroneous, a policy must first be coherent; deliberation does not guarantee coherence and under certain conditions may undermine it. The simplest version of this worry is that many minds might spin their wheels indefinitely, reaching no single answer or composite perspective at all. Theorists who combine deliberation with an epistemic conception of democracy favor the metaphor of the blind men and the elephant.51 They are optimistic that the blind men can, by pooling their perspectives through communication, arrive at an overall picture of the elephant that is better than any one of them could supply individually. In one of the oldest versions of the story, however, a raja asks the blind men what sort of thing the elephant is, and the blind men “[come] to blows” over their rival descriptions. This delights the raja, who notes that “quarreling, each to his view they cling, such folk see only one side of a thing.”52 In John Godfrey Saxe’s famous version the denouement is similar: “The disputants, I ween,/ Rail on in utter ignorance/Of what each other mean,/And prate about an Elephant/not one of them has seen!”53 Here the implicit mechanism is that deliberation further entrenches the beliefs held by differing camps, sharpening their disagreements. Moreover, one may juxtapose to the story of the blind men and the elephant the maxim that a camel is an elephant built by committee.54 If the blind men disagree about what the elephant is, they may compose their information to arrive at a superior overall view (the optimistic version of the story); they may argue interminably (in the older versions of the story); but they may also simply staple together their pictures into an incoherent whole through a process of “bargaining over beliefs.”55 To avoid interminable wrangling, the group may decide to proceed with an incoherent
51 52 53 54 55
See, e.g., Estlund, supra note 30, at 233–34. Udana 68–69: Parable of the Blind Men and the Elephant, available at http://www.cs.princeton. edu/~rywang/berkeley/258/parable.html. John Godfrey Saxe, The Blind Men and the Elephant, in The Poems of John Godfrey Saxe 135, 136 (1882). In some versions, a camel is a horse built by committee. I have lifted this excellent phrase from Robert E. Goodin and Geoffrey Brennan, Bargaining Over Beliefs, 111 Ethics 256 (2001). The use to which I put it, however, is perverse given the original context, because Goodin and Brennan want to stipulate that in cases of “bargaining over beliefs” the participants search for a “rationally grounded decision.” Id. at 260. I take this to be an additional stipulation (useful for Goodin and Brennan’s purposes), not one that is inherent in the notion of bargaining over beliefs.
M-M A
39
collective view or program.56 Such a course may well have pragmatic virtues, enabling the group to move on and get something done, but those virtues are not epistemic and it is hard to see that the pooling of perspectives from many minds has produced any positive good; rather it has created a problem that has to be pragmatically sidestepped. The largest issue here, which applies both to deliberative interactions and to the simple aggregation of beliefs or judgments without communication, involves collective incoherence: the threat that beliefs or judgments will be incoherent at the group level even if all individuals hold coherent beliefs and identical preferences.57 Even where all participants have identical preferences and talk together to decide how to promote their common good, the aggregation of beliefs and judgments58 suffers from paradoxes and impossibility results that parallel the better-known paradoxes and impossibility results that arise in the aggregation of preferences. Problems of belief aggregation and judgment aggregation have only begun to be explored relatively recently, but in the best-understood result for judgment aggregation, shifting majorities within a group can produce a kind of group-level incoherence. Consider cases in which all members of the group agree that C is correct or true only if both A and B are correct or true. The stock example is a breach of contract case in which, everyone agrees, the plaintiff wins if and only if it is true both that there was a contract and that the defendant breached. There are three judges. Under certain profiles of individual judgments, a majority of the group (say judges 1 and 2) will believe that there was a contract, while a different majority (judges 2 and 3) will believe that the contract was breached.59 The consequence is that some majority believes that each of the necessary predicates for the plaintiff ’s victory are true, and yet a majority (judges 1 and 3) believe that the defendant should win. It therefore matters whether the institutional rules aggregate 56
57
58 59
To focus the issues, I assume here that the group’s members (or a sufficient fraction of them) want to proceed with some collective view or other. In fact, this will sometimes be true, sometimes not, depending upon the group and the situation. In other cases, members can agree to disagree on the rationale for their group action, proceeding with an “incompletely theorized agreement.” See Cass R. Sunstein, Incompletely Theorized Agreements, 108 Harv. L. Rev. 1733 (1995). See Geoffrey Brennan, Collective Coherence?, 21 Int’l Rev. L. & Econ. 197 (2001); see also Lewis A. Kornhauser, Aggregate Rationality in Adjudication and Legislation, 7 Pol. Phil. & Econ. 5 (2008). I will treat belief aggregation and judgment aggregation as posing similar problems, at least as relevant to the discussion here. Here Judge 3 believes that there was no contract, but also believes that if there was one, it was breached. Although some may think that such a belief structure is puzzling, it is an ordinary case of conditional belief or belief “in the alternative.”
40
L L R
beliefs by counting votes over issues or instead by counting votes over outcomes; the collective decision will be different depending upon which approach is chosen. The important point here is that the group judgment is incoherent at the collective level even though all individuals share the same goal, to get the case right, and the same premises about the logical relationships at issue. The example involves a legal case, and in legal theory this sort of problem is sometimes called the “doctrinal paradox,”60 but the problem generalizes straightforwardly to nonlegal domains. Impossibility results parallel to those for preference aggregation have been proven for judgment aggregation. Most notably, it has been shown that “if the domain of admissible individual sets of judgments is unrestricted . . . there exists no procedure for aggregating individual sets of judgments in this domain into collective ones in accordance with a set of minimal conditions similar in spirit to those proposed by Arrow.”61 Results such as this must be treated with caution, on both the normative and positive levels. Normatively, that it is impossible to guarantee coherent aggregation of either preferences or judgments may either be good or bad, under given conditions.62 Incoherence is not the worst thing in the world; it may often be better than bad coherence. But incoherence cannot be defended on epistemic grounds, and where it obtains the Aristotelian account of the interaction of many minds seems starry-eyed. As a positive matter, for preference aggregation, it is well known that certain combinations of issues and preference profiles (single-dimensional issues and single-peaked preferences) preclude cycling, and some argue that deliberation can itself induce those conditions to hold.63 It is ultimately an empirical question whether the conditions that produce preference cycling obtain in real-world institutions; a convincing study suggests that preference cycling is not a major problem in Congress.64 60 61
62
63
64
Lewis A. Kornhauser and Lawrence G. Sager, The One and the Many: Adjudication in Collegial Courts, 81 Calif. L. Rev. 1, 10–12 (1993). Christian List, Substantive and Meta-Agreement, in Deliberation and Decision 119, 126 (Anne Van Aaken, Christian List, and Christoph Luetge eds., 2004); for the basic result, see Christian List and Philip Pettit, Aggregating Sets of Judgments: An Impossibility Result, 18 Econ. & Phil. 89 (2002). For the argument that preference cycling has beneficial effects in a democracy, see Nicholas R. Miller, Pluralism and Social Choice, 77 Am. Pol. Sci. Rev. 734 (1983); Anthony McGann, The Logic of Democracy (2006). Kornhauser, supra note 57, argues that it is too demanding to expect legislative and judicial institutions to produce aggregate rationality of judgments. See David Miller, Deliberative Democracy and Social Choice, 40 Pol. Stud. 54 (1992) (special issue); John S. Dryzek and Christian List, Social Choice Theory and Deliberative Democracy: A Reconciliation, 33 Brit. J. Pol. Sci. 1 (2003). For a critique, see McGann, id., at 130–33. See Gerry Mackie, Democracy Defended (2003).
M-M A
41
Likewise, it is possible that deliberation can induce profiles of belief that are analogous to single-peakedness for preferences, and that have a similar effect in ensuring the coherence of group-level judgments.65 All this is an open question; it remains to be seen where, and under what conditions, the problems of collectively incoherent beliefs or judgments obtain (assuming these are problems), and how deliberation affects them. What the recent results about incoherent group judgments add, however, is a hurdle that optimistic accounts of the pooling of perspectives must surmount.
Many Minds and Law: General Problems How robust are many-minds arguments? To be well specified and plausible, many-minds arguments must pass through a series of conceptual, methodological, and institutional filters. Whether they can do so depends on the particulars of the case, but it is likely that most such arguments will fail at one stage or another. I will describe the filters seriatim. The general theme is that many-minds arguments are fragile, resting on shaky mechanisms and requiring ideal conditions to survive.
Whose Minds? On Multiplicands and Denominators A well-known conceptual problem with arguments for and against majority rule involves the following problem: over what group is the majority defined? A majority of what? A “majority” is a multiplier, and the attractiveness of majority rule in particular domains or cases will often vary with different accounts of the multiplicand. The abstract arguments for “majority rule” are not equally persuasive for a majority of all legislators, of all legislators who choose to vote, of all registered voters, of actual voters, of jurors, of judges, of oligarchs, and so on. There are parallel problems with many-minds arguments in their various forms. Whose minds, exactly, are at issue? If “many minds who think X” is the numerator; what is the denominator, the set out of which the many minds are drawn? The appeal of the particular many-minds argument at issue varies as the denominator varies.66 The parallel is clearest in legal applications of the Condorcet Jury Theorem, which can 65
66
See Christian List, Group Deliberation and the Revision of Judgments: A Social-Choice-Theoretic Perspective (July 23, 2007), available at http://personal.lse.ac.uk/LIST/PDF-files/Judgment Revision.pdf; List, Substantive and Meta-Agreement, supra note 61. Ernest A. Young, Foreign Law and the Denominator Problem, 119 Harv. L. Rev. 148 (2005).
42
L L R
(but need not) be interpreted as a theorem about majority voting. Many applications of the Jury Theorem encounter problems with the multiplicand or denominator. Consider the important claim that, on Jury Theorem grounds, American courts might be more likely to make good law, somehow defined, by taking into account the law of foreign jurisdictions.67 Here foreign jurisdictions, especially courts, are understood as a virtual voting group or statistical group that will tend to converge on right answers, given the Jury Theorem’s other conditions. Which foreign courts count, however?68 Should Zimbabwe and North Korea be excluded, because their courts are too subject to political influences and other distortions, which tends to reduce their epistemic competence? Or should they be included precisely in order to reduce the correlation of biases within the virtual voting group, which might well improve overall epistemic competence?69 Here the problem, although real, is not insuperable; a suitably specified decision rule can solve it, in principle, by trading off the benefits of increased competence against the costs of increased correlation, and more broadly by maximizing the epistemic power of the notional group whose (majority) view is to be taken into account. In other examples, however, the denominator problem bites harder. Consider the argument that if, and because, there is a single right answer to all legal questions, then judicial panels should not use supermajority rules when deciding the merits of cases. After all, the Condorcetian logic is that majority voting is most likely to get the right answer in such cases, assuming the other conditions are met. But this argument begs the question: a majority of what group? If the issue in the case is whether a statute is constitutional, the legislators’ (implicit) votes that it is constitutional should enter into the denominator, at least if we assume that legislators voted on the basis of a good-faith judgment of constitutionality (by parity with our assumption that judges will decide in good faith). If so, then it is not sufficient that a majority of the judges find the statute unconstitutional. To the contrary, a supermajority rule weighted in favor of the statute’s constitutionality—such as a requirement that six of nine justices of the Supreme Court must vote to invalidate—might itself be indicated on Condorcetian grounds, as a means of giving effect to the views of many legislative minds. We can also put this point in terms 67 68 69
The best treatment is Eric A. Posner and Cass R. Sunstein, The Law of Other States, 59 Stan. L. Rev. 131 (2006). See Posner and Sunstein, supra note 67, at 158–60; Young, supra note 66, at 161–63. See Ladha, supra note 28; Ladha and Miller, supra note 28.
M-M A
43
of a many-to-many epistemic comparison between legislatures and courts, a theme I take up below. A version of the denominator problem arises in other types of many-minds arguments, particularly the Hayekian and Burkean variants. Chapters 2 and 3 will consider the idea that tradition should be a source for rules of constitutional law, because tradition embodies the contribution of many minds. As we will see, it is a major puzzle just to specify what exactly the argument suggests. Which traditions exactly, emanating from whose minds? Sometimes the implicit idea is that constitutional law should look to precedent, on the one hand, or custom, on the other. And many discussions run together these two things, treating precedent as a kind of judicial custom. But these ideas are very different.70 Precedents come, proximately anyway, from the minds of judges, whereas customs come from the minds of nonjudicial officials or the citizenry at large. Depending upon the setting and the argument, these differences can be decisive. The distinction between precedent, on the one hand, and social customs, on the other, will matter very much to neo-Hayekians, who are generally suspicious of “social engineering” by centralized government institutions, emphatically including courts. Thus some Hayekians, who favor widespread social customs and traditions as the products of decentralized “spontaneous order,” are suspicious of Supreme Court precedent on constitutional matters; they see the Supreme Court as just another centralized decisionmaker, a body of social planners in robes.71 To be sure, the justices benefit from a supporting informational network of assistants, lawyers, amici curiae, and so on, which expands the number of minds at issue. But only the justices themselves can ultimately decide, with their few minds, whether and when to accept the views of the many minds who counsel them. This unavoidable decision by few minds whether to accept the counsel of many minds forms a kind of epistemic bottleneck, a point to which I return below. Moreover, if we expand the lens this way, we must also expand the lens in similar ways when comparing the epistemic competence of the justices with that of other institutional actors, say legislators, who enjoy elaborate informational support from staff, intralegislative agencies such as the Congressional Research Service, and interest groups. Our institutional comparisons must be methodologically consistent, another point I will revisit. 70 71
John Hasnas, Hayek, The Common Law, and Fluid Drive, 1 N.Y.U. J.L. & Liberty 79, 94 (2005). A.C. Pritchard and Todd J. Zywicki, Finding the Constitution: An Economic Analysis of Tradition’s Role in Constitutional Interpretation, 77 N.C. L. Rev. 409, 489–93 (1999).
44
L L R
The most important point is just that specifying whose minds, exactly, are at issue matters a great deal. There are potentially critical distinctions between (1) the minds of many participants in social customs and traditions, (2) the somewhat fewer minds of ordinary common-law judges, and (3) the even fewer minds of Supreme Court justices developing constitutional precedent over time. Many-minds arguments in legal theory, particularly arguments for developing constitutional law through common-law methods, sometimes gain rhetorical traction by running together (1), (2), and (3) with considerable loss of precision.
More Minds, Worse Minds A basic problem for many-minds arguments is that, under identifiable conditions, more minds can be worse minds—not by happenstance, but because an increase in the number of minds itself causes a reduction in the quality of each mind. An uninteresting version of this arises under the Jury Theorem when average individual competence is less than .5 (in the binary case); when this is so, adding more members to the group will drive group competence downward toward zero. Likewise, there is a tradeoff between numerosity and average competence,72 in the sense that large numbers of minds can more than compensate for lower individual competence, so long as that competence is on average better than random. But this is an uninteresting case because the average individual competence is still given exogenously. What gives the problem a sharp edge is the possibility that competence might be endogenous to numbers. Increasing the number of minds might have collateral causal effects that systematically make minds worse. Where this endogenous effect obtains, the risk is that moving from few to many minds creates gains on one margin but losses on another. The losses may be greater than the gains or the overall effect may just be that the two margins cancel out, meaning that the institutional designer is at best running in place. I will touch on some mechanisms that can produce this sort of effect, and show that the problem is not at all confined to the Jury Theorem. Analogous problems arise for other types of many-minds arguments. In general, there are three classes of mechanisms that might reduce the group’s decisionmaking competence as numbers increase: selection 72
Bernard Grofman et al., Thirteen Theorems in Search of the Truth, 15 Theory & Decision 261, 270 (1983).
M-M A
45
effects, incentives, and emotional and social influences. (This list is slightly heterogeneous. Selection effects arise from exogenous constraints on the available pool of decisionmakers, rather than representing an endogenous effect of numbers on competence as in the latter two mechanisms. I include selection effects in this discussion because they share the feature that, where any of the three mechanisms applies, larger numbers of minds will systematically do worse). I will examine each of these mechanisms in turn.
Selection Effects The first mechanism involves constraints on selection. Madison observed that “the larger the number [of representatives], the greater will be the proportion of members of limited information and of weak capacities.”73 Here Madison was echoing, and perhaps even following,74 a famous observation of Condorcet, who believed that as the number of minds increases average competence decreases, at least in elected legislatures: A very numerous assembly cannot be composed of very enlightened men. It is even probable that those comprising this assembly will on many matters combine great ignorance with many prejudices. Thus there will be a great number of questions on which the probability of the truth of each voter will be below half. It follows that the more numerous the assembly, the more it will be exposed to the risk of making false decisions.75 There is nothing in this passage about the incentives of each to take account of the information held by others, nor is there any suggestion that a more numerous assembly will be more prone to gusts of collective emotion or rhetorical manipulation of some by others. Rather the suggestion seems to be that epistemic competence is an extremely scarce resource. It just turns out to be hard to select a large legislature whose average competence will be greater than half; beginning with the most competent, we must go further and further into the barrel to fill up a large legislature, and
73 74
75
The Federalist No. 58, at 360 (James Madison) (Clinton Rossiter ed., 1961). On historical connections, and perhaps influence, between Condorcet and Madison, see Nicholas Quinn Rosenkranz, Condorcet and the Constitution: A Response to The Law of Other States, 59 Stan. L. Rev. 1281 (2007). Condorcet: Selected Writings, supra note 11, at 49 (K. Baker ed., Bobbs-Merrill 1976).
46
L L R
by the time we are done, average competence has fallen below the critical threshold. The same claim might be made about other institutions, such as courts. Montesquieu remarked, of a Parisian court, that “decisions go by majority vote, but it is said that experience has shown that it would be better to follow the minority opinion. Which is natural enough, for there are very few good minds, and everyone agrees that there is an infinite number of bad ones.”76 There is a double irony here. Even if one might prefer the minority’s view because epistemic competence is sharply limited, it is paradoxical to infer that epistemic competence is sharply limited from the fact that everyone thinks so.
Incentives and Epistemic Free-Riding The second way in which an increase in numbers can make epistemic outcomes worse is through incentives, rather than selection effects. Bentham noted that although “the probability of wisdom increases with the number of members [of an assembly]”—the basic Condorcetian insight—there is a countervailing effect as well: “[t]he greater the number of voters the less the weight and the value of each vote, the less its price in the eyes of the voter, and the less of an incentive he has in assuring that it conforms to the true end and even in casting it at all.”77 This is familiar from democratic theory as the problem of rational ignorance; it applies not only in the setting of mass elections, but in any setting where the Jury Theorem might otherwise apply.78 A related phenomenon is the information cascade, in which individuals rationally allow the presumed information of others to swamp their private judgments.79 The information cascade can be restated as the Burkean paradox: judges who exercise their independent reason make the best contributions to the stream of precedent or tradition, while judges who follow precedent or tradition contribute nothing to its epistemic value by their decisions. I expand upon these ideas in Chapter 2.
76 77 78
79
Charles de Secondat Montesquieu, Persian Letters 167 (C.J. Betts trans., Penguin 1973) (1721). Quoted in Jon Elster, Explaining Social Behavior 413 (2007). For the basic tradeoff between group size and free-riding on the information of others, see D. Karotkin and J. Paroush, Optimum Committee Size: Quality-versus-quantity dilemma, 20 Soc. Choice & Welfare 429 (2003); Christian List and Philip Pettit, An Epistemic FreeRiding Problem?, in Karl Popper: A Critical Appraisal 128, 138–40 (Philip Catton and Graham Macdonald eds., 2004). For an overview of these phenomena, see Sushil Bikhchandani, et al., Learning from the Behavior of Others: Conformity, Fads, and Informational Cascades, 12 J. Econ Persp. 151 (1998).
M-M A
47
The import of these problems is that increasing numbers of voters rationally free-ride on the informational competence of other voters, thus decreasing average competence and undermining the jury-theoretic hope that given better-than-random competence, sheer numbers will tell. On particular assumptions, of course, these problems are not insuperable, because even tiny amounts of information held by very many individuals will aggregate to accuracy;80 on especially optimistic assumptions, rationally ignorant voters might simply abstain, in which case informed voters will dominate the turnout. But the epistemic free-riding problem is chronic and built into the very structure of the Jury Theorem. Similar problems of collective action can arise in deliberative settings in which the focus is on Aristotelian pooling of perspectives. Here the problem is free-riding on the sincerity of others. Each deliberator may speak strategically, taking the sincerity of others as a parameter while hoping to nudge outcomes in the direction of his own interests. The others, however, may of course reason likewise. As against Aristotle’s metaphor in which “a feast to which many contribute is better than a dinner provided out of a single purse,” many diners who know they will split the check pro rata will often order more than they would if paying the full price, hoping to externalize part of the cost onto others and perhaps failing to realize that others will do the same. And if they do realize that others will do the same, so that insincerity becomes common knowledge, insincerity is still a dominant strategy: whatever the others do, each is made best off by acting insincerely.
Emotions and Social Influences Third and finally, more minds may be worse minds because crowds tend to interact in pathological ways. On this view, increasing numbers create emotional and social influences that undermine epistemic competence or the preconditions for successful deliberation. Madison viewed the problem of legislative numbers—both in absolute terms and as a ratio of representatives to eligible voters—as a kind of optimization problem, trading off declining epistemic benefits against increasing epistemic costs as numbers increase. Increasing the number and ratio of representatives in the House of Representatives would, up to a certain point, improve deliberation, in part because representatives would hold more fine-grained
80
Minoru Kitahara and Yohei Sekiguchi, Condorcet Jury Theorem or Rational Ignorance? 4 (rev’d Jan. 31, 2006), available at http://wakame.econ.hit-u.ac.jp/~riron/Workshop/2006/ sekiguchi06.pdf.
48
L L R
information about local conditions. However, past the optimum, the epistemic costs of numbers start to tell as numerosity creates emotional and social influences: The truth is that in all cases a certain number [of representatives] at least seems to be necessary to secure the benefits of free consultation and discussion, and to guard against too easy a combination for improper purposes; as, on the other hand, the number ought at most to be kept within a certain limit, in order to avoid the confusion and intemperance of a multitude. In all very numerous assemblies, of whatever characters composed, passion never fails to wrest the scepter from reason. Had every Athenian citizen been a Socrates, every Athenian assembly would still have been a mob.81 It is not clear that the effects of assembly size on emotion are as Madison posited. With size tends to come diversity—of information, emotional proclivities, and social background—and diversity may counteract emotional and social influences, or break up informational or reputational cascades. On the other hand, everyone has experienced the intensity with which emotions and rumors are suddenly transmitted through large crowds, a phenomenon whose causes are as yet poorly understood. Although many sociologists reject James Madison’s concern,82 it is an enduring theme in the analysis of institutional decisionmaking.
A Note on Simultaneous and Sequential Decisionmaking An important institutional issue, one that is pervasive in the legal system, underlies several of the points made above. Simultaneous and sequential decisionmaking are very different; we cannot speak in any simple way of extending the Jury Theorem or Hayekian tacit information or Aristotelian deliberation over time, just as we might extend these things over space. Sequential decisionmaking in which decisions are revealed as they are made permits various forms of free-riding, informational and reputational
81 82
The Federalist No. 55 (James Madison), at 342 (Clinton Rossiter ed., 1961) (emphasis added). Following Madison, the classic treatment is Gustave LeBon, La Psychologie Des Foules (1895). As against Madison and LeBon, see Carl J. Couch, Collective Behavior: An Examination of Some Stereotypes, 15 Social Problems 310 (1968) (arguing that crowds are not systematically more emotional or less rational than individuals).
M-M A
49
cascades, strategic abstention, and bandwagon effects that may undermine the epistemic competence of many minds. The problem is that sequential decisionmaking is the ordinary condition of the legal system. Bentham observed that the simultaneity of legislative voting dampened “undue influence,” because legislators vote in ignorance of how other legislators have actually voted.83 For these reasons, Bentham recommended that legislatures use simultaneous decisionmaking. However, in modern legislatures and other decisionmaking committees, it is rare that everyone votes without knowledge of how others have voted, except where issues are so uncontroversial that voice voting or a show of hands is used. For controversial issues, practices such as formal or informal roll-call voting entail that voting occurs sequentially, with knowledge by later voters of how earlier voters have voted.84 In the United States Supreme Court, the justices vote sequentially in order of seniority. In mass elections, of course, voters invariably vote sequentially with knowledge of how others voted, and such elections are especially fertile ground for various forms of cascades, bandwagons, free-riding, and strategic abstention.85 Of special importance for legal theory is that a system of precedent is a system of sequential decisionmaking in which voters know the decisions of earlier voters. Even where there would be epistemic value in the aggregation of simultaneous legal judgments from many judges or other minds, one cannot straightforwardly infer that there would be equivalent epistemic value in precedent, seen as a system for aggregating the judgments of many minds over time. Far from representing a set of legal judgments that have stood “the test of time,” a body of precedent generated sequentially may embody reduced epistemic value. In place of the metaphor of the test of time, we might substitute another metaphor, seeing a line of precedent as a kind of judicial fad or fashion86 that does not even presumptively embody the contributions of many minds. None of this implies that sequential decisionmaking is always worse than simultaneous decisionmaking from an epistemic perspective. Quite the contrary, under some conditions it can be superior, especially if later decisionmakers acquire useful information from observing the actions of predecessors. But in the standard models of epistemic free-riding and 83 84 85 86
Jeremy Bentham, Political Tactics 106–07 (Michael James et al. eds., 1999). Marco Battaglini et al., Efficiency, Equity, and Timing of Voting Mechanisms, 101 Am. Pol. Sci. Rev. 409, 409–10 (2007). See Steven Callander, Bandwagons and Momentum in Sequential Voting, 74 Rev. Econ. Stud. 653 (2007). Cf. Cass R. Sunstein, On Academic Fads and Fashions, 99 Mich. L. Rev. 1251 (2001).
50
L L R
information cascades that we have mentioned, sequential decisionmaking is the crucial precondition for the problem. Some experimental evidence confirms the existence of the problem,87 although other evidence underscores the potential benefits of sequential decisionmaking.88 So the fair conclusion, given the state of the evidence, is that while sequential decisionmaking is not necessarily worse, it is always an issue; manyminds arguments must be sensitive to the details of the decisionmaking structure.
Epistemic Bottlenecks and Related Problems Madison also argued that “in all legislative assemblies, the greater the number composing them may be, the fewer will be the men who will in fact direct their proceedings. . . . The countenance of the government may become more democratic, but the soul that animates it will be more oligarchic.”89 As numbers increase, a nominal group of many minds will tend, in fact, to be dominated by few minds. Indeed, this is not only inevitable, but may actually be desirable for the many themselves, at least where direction by the few is necessary for the many to take any action or to form any coherent collective view. Still, the tendency for many minds to be directed by few undermines the epistemic superiority that many minds would otherwise display. Suppose that we have a well-specified many-minds argument, stating clearly whose minds are at issue; that all members of the relevant population vote or judge or deliberate or act independently; and that that there is no free-riding, Burkean paradoxes, or information cascades. Still, the superior epistemic judgments of many minds may simply be unusable by the legal system. The problem is that those judgments will at some point have to be funneled or refracted through the judgments of a much smaller group, perhaps a single mind, and this will result in a kind of epistemic bottleneck, or chokepoint. The judgments of many minds may be the input to a decisionmaking process, but if the structure of that process requires or allows few minds to accept or reject the many-minded judgment, or even just to interpret it, then the resulting decision may be little better than if the one mind had simply decided for itself, right from the start. Gold in, garbage out. 87 88 89
See S. Nageeb Ali et al., Information Aggregation in Standing and Ad Hoc Committees, 98 American Economic Review 181, 185 (2008). Battaglini et al., supra note 84. The Federalist No. 58, at 360–61 (James Madison) (Clinton Rossiter ed., 1961).
M-M A
51
Bentham first identified this problem in a discussion of the incorporation of custom into common-law decisionmaking. Suppose that customs or traditions—of the broader society, not of the judiciary—are not self-defining or self-applying; alternatively, suppose that the content of the custom is quite clear in the given case, but that judges have discretion in whether or not to incorporate the custom. The few minds of the judges may distort the custom, misconceive it, or reject it in favor of their own, quite erroneous, judgments. If customs are refracted through judicial decisions, then they need have no better epistemic credentials than do the judges themselves.90 In the limiting case—the nightmare scenario for many-minds arguments—a single judge, perhaps of lower than average competence, might be the chokepoint by virtue of the contingent politics of the judicial system. Suppose that on the high court of the jurisdiction a nine-member court is routinely split, four to four, and that one judge of low epistemic competence is routinely the swing vote. In these circumstances, a many-minds argument for the epistemic credentials of tradition, or custom, or a long line of judicial precedent, actually boils down to an argument for tradition or custom or precedent as refracted through the mind of the epistemically limited judge. That judge will in effect be the only one who interprets and applies the relevant tradition or custom or precedent; many minds will be funneled through one. The same holds for the argument that the epistemically limited judge should defer to the commands or policy choices embodied in statutes, because a large group of legislators, deliberating or aggregating their information, have reached an epistemically impressive judgment. The statute may plainly rule out some results but in hard cases, the ones that reach appellate courts, the swing judge will have a broad scope to make his own (low-competence) judgment about whether to accept the legislative judgment, and about what that judgment is best understood to mean. The resulting decision may or may not be good, but many-minds arguments will not establish whether it is; all will be reduced to the performance of a single mind or a small group. I have used an example from the judicial setting for vividness, but of course the same problem may hold, with appropriate modifications, in other institutions. Even if legislators of high average competence could
90
Cf. Jeremy Waldron, Custom Redeemed by Statute, 51 Current Legal Probs. 93, 100 (1998) (“Bentham argued that once they come into the hands of the judges, customs tend to be ill-used and subjected to arbitrary and unpredictable modifications.”).
52
L L R
pool their many minds in epistemically impressive ways, the legislative leadership may form a kind of chokepoint that prevents them from doing so; perhaps the wisdom of the legislative multitude must be approved by, or at least refracted through, the mind of a Nancy Pelosi. The numerous “vetogates” in the federal legislative process ensure that the whole group of many minds often cannot bring its group-level epistemic competence to bear as such. Rather decisions are made serially by a series of separate groups of few minds (such as committees), and it is largely beside the point that, absent this structure, many minds might have displayed higher epistemic or deliberative capacities than the few. Similar problems obtain in the executive branch. Although the executive is a “they,” not an “it,” still the hierarchical structure of the executive usually implies that at some point the decision supported by many experts or mid-level officials will be funneled upward to a chokepoint, coming to rest on the desk of a single mind who can approve, or disapprove, or modify, or interpret, before issuing a final decision. In this setting, as in the others, it is no good to say that we can just urge the one or few minds to accept the conclusions of many minds. The fact that one or few minds must unavoidably make the decision, with limited epistemic competence, whether and when to accept the counsel of many minds is precisely what constitutes the epistemic bottleneck. Epistemic bottlenecks are only one of a class of related problems that have been insufficiently explored. Machiavelli argued that “a multitude without a head is useless”;91 an undifferentiated mass of people cannot make their views known—even if they have common views—without speaking through a smaller subgroup or sole representative. Once that representative is appointed, a kind of epistemic slack arises: even if the leader shares the preferences of the group, she will inevitably filter the group’s common views or plans through her own, more limited understanding. Although this may distort the group’s views or plans, the tradeoff is that absent this distortion the group’s view may never be expressed at all and its plans may never result in action. Indeed, to coordinate on a group judgment or belief or view in the first place, many minds need an epistemic agenda-setter to help them sort between the possibilities. Just as with preference-based choices, so too with epistemic choices: decisions may be sensitive to the order in which they are presented, and where they
91
Niccolo Machiavelli, Discourses on Livy, bk. 1, ch. 44 (Harvey C. Mansfield and Nathan Tarcov trans., 1996).
M-M A
53
are, the outputs of many minds will be structured by the views of the few agenda-setters. Overall, there is an epistemic analogue to the iron law of oligarchy.92 The inevitability of epistemic chokepoints, epistemic slack, and epistemic agenda-setting makes some types of many-minds arguments seem utopian; “the wisdom of crowds” becomes “the wisdom of the chaperones.”93 Many minds inevitably require the leadership of few minds in order to form and express any views or to get things done. Where this is so, the epistemic superiority of the many is constrained by the imperatives of organization for collective action.
Many Minds vs. Many Minds: Comparative Epistemic Competence For present purposes, the most serious problem with many-minds arguments involves comparative epistemic competence across institutions. The problem arises because many-minds arguments typically compare the one or the few with the many. A main point of the Jury Theorem, for example, is that group competence can quickly exceed that of the group’s most competent single member. Hayekian or Burkean evolutionary arguments for the epistemic power of the common law compare the cumulative wisdom of many judges, over time, with the unaided reason of a single judge or small judicial panel. Aristotle’s metaphors compare the pooled perspectives of the many with the partial perspectives of the “few good.” I will call these one-many comparisons, and I will suggest that they provide little support for epistemic legalism and are of little utility for legal theory. Even if one-many comparisons of this sort succeed, the legal system typically presents a very different type of issue: many-many comparisons, in which institutions staffed by many minds are on both sides of the comparison. In such comparisons, the details will matter critically for the epistemic analysis, but we do not get much help with the details by knowing, for example, that the crowd or the multitude can be epistemically superior to a single individual or expert under certain conditions. No amount of onemany comparisons will get much purchase on complex many-many institutional comparisons that are necessary to evaluate epistemic legalism. 92 93
See Robert Michels, Political Parties (5th ed. 2007). See Chris Wilson, The Wisdom of the Chaperones: Digg, Wikipedia, and the Myth of Web 2.0 Democracy, Slate (Feb. 22, 2008). For research showing that 1 percent of Wikipedia users are responsible for about half of the site’s edits, see http://asc-parc.blogspot.com/2007/05/longtail-and-power-law-graphs-of-user.html.
54
L L R
I take up these points in detail in chapters 2 and 3, with particular reference to common-law constitutionalism and to Burke, Bentham, and Thayer. My basic suggestion is that common-law constitutionalism is most plausible when it means that precedent and tradition should trump the idiosyncratic theorizing of the individual judge, and least plausible when it means that precedent and tradition should trump the many minds of the current legislature.
Conclusion: The Limits of Many Minds I have expressed a qualified skepticism about the utility of many-minds arguments for legal theory. Although many-minds argumentation is a heterogeneous category, comprising aggregative, evolutionary, and deliberative ideas that bear only family resemblances to each other, still these different types of arguments are often treated together and there is good reason to do so. If there is a single feature common to all these arguments, it is that the one or the few cannot hope to compare with the many in point of epistemic quality. Against this general view, I have made the following suggestions: (1) “Many minds” is a slippery group, whose epistemic credentials depend on precisely who is included and excluded. (2) As the number of minds increases, the quality of minds may decrease endogenously, due to selection effects, incentives for rational ignorance and free-riding, and emotional and social influences. (3) There is an epistemic analogue to the iron law of oligarchy: many minds require the help of an epistemic agenda-setter and cannot speak or act except under the leadership of the few, which gives rise to epistemic chokepoints and epistemic slack. (4) In the legal system, one-many comparisons are rare, while manymany comparisons are ubiquitous. Overall, my claim is in the alternative: many-minds arguments are fragile and, where they are robust, have a different institutional valence than epistemic legalism suggests. Hayekians and Burkeans conflate custom and the common law, either inadvertently or in order to burnish the common law’s epistemic credentials; common-law judges distort custom even when, and because, they try to absorb it into their decisions; and common-law constitutionalism at the level of the Supreme Court makes
M-M A
55
flawed institutional comparisons, applying one-many arguments to problems that have a many-many structure. These examples show that there is no necessary or causally important connection between the epistemic virtue of many minds, on the one hand, and epistemic legalism’s commitment to judicial lawmaking, on the other. Rather, under robust conditions, many-minds arguments will favor lawmaking by legislatures and other nonjudicial institutions—or so I will go on to claim.
This page intentionally left blank
Chapter 2
The Constitutional Common Law: Information Aggregation
T
he main testing ground for epistemic legalism is the allocation of constitutional authority between courts and legislatures. How can that allocation be structured to maximize the epistemic quality of lawmaking—to maximize the extent to which laws are based upon factual, causal or moral truth, or the truth about other relevant laws, such as the Constitution? How do many-minds arguments help, or fail to help, in carrying out this task? In this chapter I examine these questions in their most crucial setting: common-law constitutionalism. Common-law constitutionalism is not a single theory but rather a family of theories about constitutional adjudication and change. As we have seen, common-law constitutionalism has several antonyms, so the family of theories has several members. It includes the ideas that courts do and should develop the meaning of vague or ambiguous constitutional texts by reference to tradition and precedent, rather than the original understanding or the judges’ unaided moral theorizing, and the related idea that courts do and should proceed in a Burkean rather than ambitiously 57
58
L L R
rationalist or innovative fashion. In recent years, the central and most striking claim of common-law constitutionalism has been that precedent and tradition embody some form of latent wisdom. Judges will generally do best by deferring to the collective wisdom embodied in precedent and tradition, rather than trusting to their unaided reason, or so the general claim runs. In this chapter and the next, I offer a critical analysis of the mechanisms that are said to generate this latent wisdom. Drawing throughout on Bentham’s critique of the subconstitutional common law, I attempt to update Bentham by using the tools of modern social science and by adapting his claims to the setting of constitutional law. My conclusions, however, remain largely Benthamite in spirit: the constitutional common law is not plausibly seen as a repository of latent wisdom. The relevant claims and mechanisms suffer from infirmities of internal logic and from a failure to make institutional comparisons between and among precedent and tradition, on the one hand, and the outputs of legislatures, executive officials, and constitutional framers, on the other. The idea most frequently invoked to connect common-law constitutionalism and the limits of reason is Burke’s Dictum: “We are afraid to put men to live and trade each on his own private stock of reason, because we suspect that this stock in each man is small, and that the individuals would do better to avail themselves of the general bank and capital of nations, and of ages. Many of our men of speculation, instead of exploding general prejudices, employ their sagacity to discover the latent wisdom which prevails in them.”94 Burke’s dictum is a claim about ordinary common law, not constitutional common law; Burke celebrated a common-law system that did not include judicial review (at least not in anything like its modern forms).95 But common-law constitutionalists, invoking Burkean themes, have transposed into the setting of constitutional law the putative mechanisms that generate the latent wisdom of tradition, emphasizing the immanent rationality and evolutionary or adaptive fitness of constitutional precedents and traditions. As we shall see, these ideas can be interpreted either in informational terms, 94 95
Edmund Burke, Reflections on the Revolution In France 183 (W. Alison Phillips and Catherine Beatrice Phillips eds., 1912) (1790). See Phillip A. Hamburger, Revolution and Judicial Review: Chief Justice Holt’s Opinion in City of London v. Wood, 94 Colum. L. Rev. 2091 (1994).
T C C L: I A
59
through the lens of the Condorcet Jury Theorem, or in evolutionary terms, by reference to models in which common-law processes evolve towards efficiency. In the latter case, Burkean ideas about the evolutionary value of tradition, custom, and precedent overlap a great deal with Hayekian ideas. Where Burkean ideas are interpreted in Condorcetian terms, they are quite distinct from Hayek’s evolutionary views. In either case, however, I will critique the mechanisms said to underpin the latent wisdom of tradition. Drawing upon the discussion in Chapter 1, I suggest a range of puzzles and problems that limit the mechanisms’ operation to a narrow set of conditions and that sharply limit their relevance for constitutional law and theory. Common-law constitutionalism may or may not be justifiable or superior to its competitors on other grounds, but the informational and evolutionary mechanisms recently invoked to depict it as a repository of latent wisdom turn out to be intrinsically fragile and institutionally ungrounded. The central theme in the following critique is this: arguments for the rationality or efficiency of the ordinary common law, or of societal traditions, do not translate successfully into arguments for the rationality or efficiency of the constitutional common law, especially as compared to statutes and other sources of law. The institutional context of constitutional adjudication is decisively different than that of ordinary commonlaw adjudication. Both the epistemic and evolutionary mechanisms at most suggest that a single judge can do better by deferring to the collective wisdom embodied in the decisions of past judges, or in larger societal traditions, rather than by relying on her unaided reason or idiosyncratic theorizing. This claim, however plausible on its own terms, turns out to have little relevance to the setting of constitutional judicial review. Perhaps the claim works well enough in the shrinking domain of ordinary common-law cases, where there is no statute or administrative regulation in the picture, although I shall question even that possibility. Yet even if the claim holds for the ordinary common law, it does not describe the situation that judges face in constitutional adjudication. The alternative to relying on precedent or tradition, in constitutional law, is rarely or never reliance on the unaided reason of the single judge; the alternative is reliance on the latent wisdom of collective legislatures, or of the executive branch, or of a group of constitutional framers. In different contexts, each of these sources has distinctive costs and benefits. However, the relevant institutional comparisons are much different than, and more complex than, the simple comparison of unaided reason to collective wisdom emphasized by Burke.
60
L L R
In short, common-law constitutionalism trades on the ambiguity of its antonyms. As against the idiosyncratic theorizing of individual judges, common-law constitutionalists argue that judges should follow precedent because many judicial heads are better than one. Although this can be true under certain conditions, it is not true in others; Chapter 1 outlined conditions under which more heads are actually worse than one. In any event, even where the product of many judicial minds is indeed better than idiosyncratic judicial theorizing, this does not at all entail that many judicial minds are better than many legislative ones. Where the antonym of common-law constitutionalism is legislative constitutionalism that settles and liquidates the meaning of vague or ambiguous provisions, and updates constitutional law over time, epistemic considerations argue for judicial deference to legislatures, or so I will claim. If the social goal is to maximize the epistemic quality of the laws, the best allocation of lawmaking authority is a regime of common-law constitutionalism by legislatures and deference by courts.
Plan of the Critique This chapter discusses Burke as Condorcet—the idea that Burke’s dictum can be interpreted epistemically, in light of the information-aggregating models developed under the umbrella of the Condorcet Jury Theorem. I suggest that the Jury Theorem has no clear payoff for constitutional adjudication. The Theorem’s rather stringent conditions will frequently be violated by the precedents or societal traditions that common-law constitutionalists would draw upon. Most importantly, justifying common-law constitutionalism on the basis of the Jury Theorem is fatally noncomparative. Statutes, administrative decisions, and constitutional texts also embody information and are also the product of many minds. In a range of cases, those sources will be superior to precedents and societal traditions, according to the Jury Theorem’s own criteria and logic. Indeed, the epistemic interpretation of Burke leads, via Condorcet, to the views of James Bradley Thayer—to deference to legislatures in constitutional adjudication. Chapter 3 discusses Burke as Darwin and Hayek—the idea that Burke’s dictum can be interpreted in light of evolutionary models, reminiscent of Hayekian ideas, in which the common law converges to efficiency, as if by the operation of an invisible hand. I suggest that these models, like the Jury Theorem, have no obvious payoff for constitutional adjudication. Even if efficiency or adaptive fitness is a desirable property of the ordinary common law, it is a problematic normative goal for constitutional law.
T C C L: I A
61
In any event, the mechanisms of common-law evolution have no clear counterpart in constitutional law. The central evolutionary mechanism in common-law theory—that the selection effects of litigation cause inefficient rules to be weeded out over time—is of dubious relevance to constitutional law, where litigation is infrequent and stakes are chronically distributed in an unequal way across organized and unorganized interests. Most important is that even if the constitutional common law is constantly converging to efficiency, the background environment is changing as well, which means that constitutional precedent may or may not be efficient at any given time. If the political environment changes with sufficient rapidity, constitutional common law will face a moving target, and will not have enough time to converge to or even close to efficiency. For that reason lawmaking through alternative procedures, such as legislation, will often prove at least as efficient as precedent. Even if constitutional law is constantly converging towards the optimum, legislation may be closer to the optimum at any given point in time, given a changing environment. Here too the crucial institutional comparison is between common-law constitutionalism entrusted to legislatures, on the one hand, and common-law constitutionalism entrusted to judges on the other. There is no general reason to think that the evolutionary capacity of the judicial version is systematically superior to that of the legislative version. Even if the efficiency thesis is useful in the ordinary subconstitutional context of common-law adjudication, its utility is obscure in the very different setting of constitutional law. Throughout, I pursue a two-fold strategy, both identifying internal problems with the optimistic accounts of the common law that I canvass, and then proceeding to examine whether those accounts, even if correct, can be transposed to the institutional setting of constitutional adjudication. In each case, the internal critique is a necessary preliminary to the constitutional discussion; the mechanisms of common-law constitutionalism that I discuss cannot be understood in the abstract, without reference to the common-law context from which they arose. While I am skeptical that these mechanisms offer successful accounts of the ordinary common law, my central claim is that even if they do, no transposition to the constitutional setting is possible. All in all, the idea that common-law constitutionalism is a repository of latent wisdom, and enables judges to cope with the limits of human reason, finds little support in informational or evolutionary mechanisms. This does not touch arguments that justify common-law constitutionalism on other grounds, such as the value of stability and protecting settled
62
L L R
expectations, nor does it speak to other debates over constitutional adjudication. It does, however, remove a main strut of epistemic legalism, for it turns out that common-law constitutionalism, in its judicial version, cannot be justified by reference to the limits of reason.
Burke as Condorcet I will begin by explaining, and questioning, the informational version of common-law constitutionalism (by judges; where the qualifier is not explicitly stated it should be understood). I first outline some basics of the Burkean approach to constitutional adjudication, attempting to specify the Burkean claims more precisely and to identify the many puzzles that they pose. I then detail a Condorcetian interpretation of Burke’s dictum as a mechanism for aggregating information: the decisions of a line of past judges embody aggregated information that is superior to the unaided reason of any single judge or small group of judges sitting today. I proceed to analyze a series of problems arising from the internal logic of this interpretation, suggesting that the Jury Theorem’s conditions are too restrictive for the interpretation to be convincing. Finally, I turn to comparative institutional problems. The idea that decisions of a line of past judges are better than the decisions of a single judge today has little relevance for constitutional adjudication. There the question is not whether judges should be guided by their unaided reason; it is whether they should defer to past judges rather than the views of constitutional framers or current legislative and executive institutions. In those comparisons, the Condorcetian logic suggests that in many cases, precedent is not the best source of information. Among the alternatives, I shall argue that current legislation is the Condorcet winner, to adapt a phrase. The original understandings of the founding generation are hopelessly outdated and thus fail to use crucial information, whatever their other epistemic virtues, whereas statutes enjoy a range of Condorcetian advantages.
Burkean Basics What exactly is the Burkean view, of law generally and of constitutional adjudication in particular? It turns out that there is no one view; Burkeanism has multiple strands, intersecting in complex ways. Nonetheless, there is a core Burkean commitment to deriving from the limits of reason a robust role for the common law, both ordinary and constitutional, and hence a robust role for judicial lawmaking.
T C C L: I A
63
Common-law constitutionalism represents a self-consciously Burkean strain in constitutional theory.96 The basic commitment of Burkean theorists is to tradition, somehow understood. For some, tradition has intrinsic worth, and adherence to tradition is intrinsically admirable or at least inevitable. On this view, tradition is constitutive of our very identities.97 To break with tradition threatens a kind of cultural suicide, and in any event is necessarily ineffective, because tradition will continue to shape the very attempt to break with it. For other Burkeans, however, tradition has instrumental or derivative value.98 Tradition can be and has been defended on a variety of instrumental grounds, most prominently on epistemic grounds. These include the idea that tradition aggregates information from many minds and the idea that institutions evolving incrementally over time are more likely to be optimal than designed institutions, which fail due to the limits of the designers’ reason. I examine the latter idea in Chapter 3. This chapter addresses the argument from information aggregation. Taken together, these epistemic arguments for Burkeanism are enduring and central; they unite many disparate strands of Burkeanism in a common distaste for “theorizing,” “unaided reason,” “rationalism,” and “innovation.” In a more elevated vein, Burkeans sometimes suggest that tradition is the collective analogue of individual wisdom. At the individual level, wisdom is an attribute that is notoriously hard to define. Fortunately, I need not do so here. What matters for my purposes is just that Burkeans depict tradition at the level of society as analogous to wisdom at the level of the individual. Just as the individual acquires wisdom (if one is lucky) over the course of a life, so too societies acquire traditions over time, and these traditions embody the wisdom of the many minds who have made incremental contributions to them. As will see, however, this attempt to transpose wisdom in the individual case to tradition at the collective level turns out to be fallacious.
Tradition or Precedent? Or Both? Burkeans who apply their views to adjudication tend to assimilate precedent and tradition; they see common-law constitutionalism as a kind of 96
97 98
See Thomas W. Merrill, Bork v. Burke, 19 Harv. J.L. & Pub. Pol’y 509 (1996); Cass R. Sunstein, Burkean Minimalism, 105 Mich. L. Rev. 353 (2006); David A. Strauss, Common Law Constitutional Interpretation, 63 U. Chi. L. Rev. 877 (1996); Ernest Young, Rediscovering Conservatism: Burkean Political Theory and Constitutional Interpretation, 72 N.C. L. Rev. 619 (1994). Anthony Kronman, The Lost Lawyer: Failing Ideals of the Legal Profession (1995). See, e.g., Merrill, supra note 96, at 515.
64
L L R
judicial tradition unfolding over time, one that draws on larger social and political traditions as well. To be sure, in some cases, or at some times, common-law constitutionalists distinguish precedent from tradition. As we will also see, however, there is also a standing temptation to let the distinction erode. The slippage from tradition to precedent and back again is a material ambiguity, indeed a crucial one. Arguments that find latent wisdom in tradition may not translate successfully to precedent, even if precedent is understood as judicial tradition. My focus is on precedent, not on tradition per se, but one of my points is that common-law constitutionalism offers similar rationales for incorporating both into constitutional adjudication, and not infrequently equivocates between the two. Accordingly, I will generally refer, with some awkwardness, to “precedent or tradition,” but will distinguish the two when the analysis warrants.
Burkeanism and Second-Order Rationality By a kind of mental commitment, the Burkean judge suspends his or her own first-order reason about the merits of individual policies in order to follow a simpler second-order rule of thumb: policies that comport with the stream of precedent or tradition, somehow defined, are permissible, at least presumptively; policies that do not are not permissible, or at least are suspect. The consequentialist argument for this decision procedure is that over an array of future cases, the Burkean judge believes she will do better by following precedent or tradition than by following her unaided reason. The Burkean approach is thus a kind of second-order rationalism that distrusts first-order rationalism, in which judges deciding particular cases attempt, on a blank slate, to figure out for themselves what is best.99 The best strategy of all would simply be to combine one’s current information with whatever information precedent or tradition contain, giving all sources of information their due weight. The Burkean view itself assumes that imperfect decisionmakers cannot do so; they are constrained to substitute tradition for independent reason, to a greater or less extent, rather than simply adding the two together. I follow that assumption in order to engage with Burkeanism’s internal logic. The shape of this argument suggests four main questions. First, is this sort of mental commitment psychologically feasible? Second, what is the criterion for “doing better”—what is the first-best that the Burkean 99
See, e.g., Strauss, supra note 96, at 892–94.
T C C L: I A
65
second-order approach hopes to track, over an array of cases? Third, why is precedent the right second-order guidepost, as opposed to, say, the findings of law and economics, the platform of the Republican Party, or the teachings of Zoroaster? Finally, what are the implications of the Burkean approach for constitutional adjudication? I take up the last question in detail below. Here I offer some brief remarks on the first three questions. The mental commitment to a second-order rule—in this case, following precedent or tradition—may seem psychologically mysterious. Suppose a case where, to the decisionmaker’s unaided reason, the correct result seems clearly to be X, whereas precedent or tradition clearly says Y; will not the second-order commitment break down? However, commonlaw constitutionalists implicitly assume that such mental commitments are stable, although the psychological mechanisms that support them are as yet poorly understood; I will follow that assumption in order to engage the internal logic of the common-law constitutionalists’ view. Moreover, two wrinkles dilute the force of the problem. First, Burkean judges use precedent or tradition not only to avoid error but to conserve on the costs of decisionmaking.100 If so, then the Burkean judge will not (re)decide the first-order question and will not form an independent view that X is the case; no tension will arise. As we will see, however, while this wrinkle helps to sidestep the problems with second-order decisionmaking, it actually undercuts the Burkean assumption that precedent or tradition contains latent wisdom, because it means that most precedents or traditions do not embody any independent judgment about the merits. I take up this crucial point below. Second, some theorists envision Burkean adjudication not as an iron-clad commitment, but as a presumption or a rule with exceptions, and thus as defeasible in particular cases where the Burkean judge’s firstorder views or judgments are particularly strong. “[A] rationalistic account of traditionalism just establishes a requirement that one give the benefit of the doubt to past practices. If one is quite confident that a practice is wrong . . . this conception of traditionalism permits the practice to be eroded or even discarded.”101 Because beliefs have a dimension of intensity, or the epistemic confidence the decisionmaker attaches to them, the Burkean can reconcile his first-order and second-order beliefs by saying that the latter will prevail only if the former do not exceed some threshold of intensity. 100 101
See id. at 912. See id.
66
L L R
A more troublesome question about the second-order interpretation of Burkeanism is: what exactly is the first-order good that following precedent or tradition promotes? If precedent or tradition is of derivative value, as an epistemic aid, of what value is it derivative? But it is wrong to assume that there must be some single first-level answer. If Burkeanism is a second-order epistemic strategy, it can be used by decisionmakers with differing first-order commitments.102 Burkeans with various first-order theories about what makes outcomes good, or valuable, or just, can converge on the second-order value of precedent or tradition, without settling their theoretical differences. It is perfectly coherent for a judge who believes that maximizing “welfare” is the touchstone of good decisions and a judge who believes that respecting “justice” is the touchstone to agree on the second-order epistemic value of precedent or tradition, as a guide to implementing their very different first-order values. The most basic question in such cases, however, is why the secondorder guidepost should be precedent or tradition, rather than something else. For Burkeans, one prominent answer is that following precedent and tradition best comports with the limits of human reason; precedent and tradition embody the contributions of many minds in the past, in contrast to the unaided reason of an individual decisionmaker today.103 Again, the many minds might be conceived as expressing views about welfare, or justice, or anything else; the basic idea that more heads are better than one or a few will remain unaffected.
Precedent, Tradition, and the Jury Theorem The Jury Theorem, introduced in Chapter 1, provides the most obvious framework for an informational interpretation of Burke’s dictum, and it is the only well-specified framework that appears in the literature on common-law constitutionalism. Common-law constitutionalists either invoke the Theorem to put structure into their claims or else remain vague about the informational mechanisms they have in mind.104 In this chapter I will therefore focus on the Jury Theorem, but many of my points will hold as against any informational account of legal Burkeanism. 102 103 104
Cf. Cass R. Sunstein, Incompletely Theorized Agreements, 108 Harv. L. Rev. 1733 (1995). See Merrill, supra note 96, at 519–23; Strauss, supra note 96, at 891–92; Sunstein, supra note 96, at 370; Young, supra note 96, at 644–47. For the Condorcetian interpretation of Burke, see Sunstein, supra note 96, at 370–71. Something like the Condorcetian interpretation is plausibly implicit in the other accounts cited in note 96, supra, but those discussions are vague about their epistemic assumptions.
T C C L: I A
67
How exactly does the Jury Theorem relate to Burkean traditionalism? On this view, traditions—including judicial traditions or lines of precedents—are seen as a series of “votes” that aggregate to a collective view. That view contains a kind of latent wisdom, in that the collective view is under certain conditions much more likely to be correct than the view of any individual. “If countless people have committed themselves to certain practices [over time], then it is indeed probable that ‘latent wisdom’ will ‘prevail in them,’ at least if most of the relevant people are more likely to be right than wrong.”105 On this account, the people who participate in a tradition or the judges who participate in developing a line of precedent lived at different times and thus never participated in an actual collective vote. The Jury Theorem does not literally require that the decisionmaking group ever take a collective vote under majority rule; exactly the same aggregative properties can be obtained just by taking the statistical mean of guesses from within some population of guessers. As the number of participants becomes large, the median vote that prevails under majority rule will converge towards the statistical mean of guesses within the group. Later I will question whether this account succeeds, but for now I take it as given. As throughout the book, I will assume that there is a truth that legal institutions can track—that there are exogenously defined right answers to the relevant legal questions. Jurisprudential controversies abound here, between those who argue that there are right answers even in hard cases,106 those who deny this,107 and those who think that the existence of right answers are irrelevant, because of persistent disagreement over what the right answers are.108 The epistemic project assumes that the thoroughly skeptical positions in this debate are wrong; if the discussion is to be worthwhile, there must be at least some nontrivial domain in which there are right answers to legal questions, answers that can at least sometimes be demonstrated as such to everyone’s satisfaction. Although this is an assumption of any epistemic discussion, I will underscore that it is also a substantively plausible assumption in this context, because of the capaciousness of the Jury Theorem. Recall the various interpretations of the Jury Theorem discussed in Chapter 1. In the 105 106 107
108
Sunstein, supra note 96, at 371. Ronald Dworkin, Hard Cases, 88 Harv. L. Rev. 1057 (1976). For an overview of different versions of the basic thesis that law is indeterminate, at least in some cases, see Lawrence B. Solum, Indeterminacy, in A Companion to the Philosophy of Law And Legal Theory 488 (Dennis Patterson ed., 1999). Cf. Jeremy Waldron, The Irrelevance of Moral Objectivity, in Natural Law Theory: Contemporary Essays (Robert George ed., 1992).
68
L L R
informational interpretation of the Jury Theorem, as opposed to the polling interpretation, the Theorem requires an exogenous or preferenceindependent right answer, and thus gets purchase on any case or legal problem where one of the following conditions are met: (1) there is a factual component to the legal question; (2) there is a prescriptive or means-end judgment about which legal ruling will best conduce to achieving an agreed-upon goal; (3) the legal question, although neither factual nor prescriptive, otherwise has a right answer somehow defined. Few will deny that many cases fit one of these categories. Even those who deny that all legal cases have right answers rarely deny that some cases do. Moreover, we must also consider the polling interpretation of the Jury Theorem—the case where (4) there is a right answer about the preferences of some larger group from which the decisionmakers are drawn. This case arises when legislators or judges decide a legal question by asking what citizens want about a given issue, or when the question is whether legislators or judges are more likely correct about what answer will best promote the majority’s interests, as in Bentham’s claim that legislators are more likely than judges to understand the common reason of the people. In such cases, the Jury Theorem also applies, given its other conditions, even though there is no right answer independent of people’s preferences.
Burke as Condorcet: Internal Problems I will now proceed to my main critical business: to show that there are serious obstacles to any Condorcetian interpretation of Burke, both in general and as applied to judicial precedent in particular. I begin with the internal logic of the interpretation, before moving to comparative institutional problems.
Precedent, Tradition, and Numbers As mentioned above, a central ambiguity within common-law constitutionalism is the slippage back and forth between judicial precedent on the one hand, and broader societal traditions, on the other. In principle, a precedent need not draw upon tradition, and a tradition need not be embodied in a precedent. Sometimes common-law constitutionalists explicitly distinguish precedent from tradition.109 Some proponents of 109
See David A Strauss, Tradition, Precedent, and Justice Scalia, 12 Cardozo L. Rev. 1699, 1706 (1991).
T C C L: I A
69
tradition, especially those who see tradition as a kind of spontaneous order, explicitly deny that praise for tradition implies praise for precedent, which they see as the positive lawmaking of a centralized lawmaker, such as the Supreme Court.110 However, the lines frequently blur, both in theory and in practice. In theory, a main claim of common-law constitutionalism is that a stream of precedent that has stood the test of time is a kind of tradition, and deserves respect for the same reasons as tradition.111 In practice, it has been argued, for example, that the Warren Court was a “common-law court,” basing its decisions on experience.112 In some cases, this claim is supported by pointing to earlier precedents on which the Warren Court drew.113 Where the Warren Court overruled or broke dramatically from precedent, however, the claim is supported by pointing to larger political and social traditions said to be inconsistent with the discarded precedent. Thus the innovative decisions requiring a “one person, one vote” standard in reapportionment cases are justified by pointing to a broader historical trend toward expansion of the formal franchise.114 Common-law constitutionalism that ranges over both precedent and tradition has multiple degrees of freedom. This ambiguity is often material, because the Condorcetian credentials of precedent may be much different than those of broader traditions. In the case of Supreme Court precedents, for example, only 110 justices have ever sat on the Court, and of those only a fraction have been participants in any given line of precedent. For a Condorcetian analysis, numbers are critical. When the individual reason of a lone judge or small group of judges is compared with traditions embodying (let us suppose) thousands or even millions of individual judgments, the superiority of collective wisdom is palpable. When the judgments of a group of nine justices, today, are compared with the judgments of (say) twenty justices yesterday, the margin of superiority thins. Moreover, by the Jury Theorem’s terms, group competence is a function not only of numbers but of average competence—the probability of selecting the correct answer. If today’s justices are more competent than yesterday’s, the difference may swamp 110 111 112
113 114
See A.C. Pritchard and Todd J. Zywicki, Finding the Constitution: An Economic Analysis of Tradition’s Role in Constitutional Interpretation, 77 N.C. L. Rev. 409, 489–93 (1999). See, e.g., Strauss, supra note 96, at 892. David A. Strauss, The Common Law Genius of the Warren Court (Univ. of Chi. Pub. Law & Legal Theory Research Paper Series, Paper No. 25, 2002), available at http:// papers.ssrn.com/ sol3/papers.cfm?abstract_id=315682. See id. at 17–24. Id. at 31.
70
L L R
a small deficit in numbers. Shortly, I will give reasons to think that average competence indeed rises over time. Of course, even if that is true, the relative numbers of yesterday’s justices may swamp the higher competence of today’s justices; everything depends on the particular values of the variables, as is always true with the Jury Theorem. The overall point remains, however. Precedent, especially at the level of the Supreme Court, has a very different Condorcetian status than large-scale political and social traditions. Assimilating the two or vacillating between them, as often happens in common-law constitutionalism, gives precedent an unwarranted sheen. To the extent that the informational interpretation of Burke is focused on judicial traditions—the precedents developed within courts, as opposed to broader social and political traditions—its Condorcetian credentials are especially suspect.
Tradition, Time, and Information I have suggested that Burkeans see tradition as the collective analogue of wisdom at the level of individuals. As Bentham and Pascal in effect argued, however, this move rests on a fallacy of composition.115 Society is not like a big person who ages over time and (if lucky) acquires wisdom. As Bentham wrote, “[a]s between individual and individual living at the same time and in the same situation, he who is old possesses, as such, more experience than he who is young. But, as between generation and generation, the reverse of this is true.”116 The generation of today is wiser than the generations of yesterday in the sense that they have more experience and information; the current generation benefits by being able to see how the plans, projects, and experiments of past generations actually turned out. At the social level, traditions developed long ago embody, not the wisdom of old age, but rather “the wisdom of the cradle.”117 In Condorcetian terms, the point entails that there is a systematic reason for thinking that today’s political actors, including judges, are of higher average competence than yesterday’s, all else equal: today’s political actors, including judges, know more, just because they live today, not yesterday.
115
116 117
Jeremy Bentham, The Handbook of Political Fallacies 43–53 (Harold Larrabee ed., 1952); Blaise Pascal, Preface to the Treatise on Vacuum, in Thoughts, Letters, and Minor Works 444, 449 (Charles W. Eliot ed., M.L. Booth et al. trans., 1910). Bentham, id. at 44. Id. at 45.
T C C L: I A
71
Common Questions? A fundamental assumption of the Jury Theorem, insofar as the aggregation of judgments is concerned,118 is that voters or group members must be addressing the same question.119 If A makes a judgment about question X, and B makes a judgment about question Y, there is no genuine pooling of collective wisdom. Suppose that X and Y superficially resemble each other, but are critically different on closer inspection. Then it may be easy to mistake an ersatz agreement among many minds, which looks as though it rests on collective wisdom but really does not, for a genuinely Condorcetian process. This ersatz collective wisdom is, plausibly, the ordinary state of judicial precedents extended over time.120 Suppose that at Time 1 the justices of the Supreme Court hold that a state may not constitutionally ban flag-burning121 or partial-birth abortion.122 Then suppose that Congress enacts a similar statute,123 and the constitutionality of that statute is called into question at Time 2. Does the collective judicial judgment at Time 1 contain information useful at Time 2, from a Condorcetian perspective? What question exactly are the justices addressing in the two cases? It is not obvious that the considerations are the same. Holmes, among others, thought that the problem of judicial review was very different for federal and state governments;124 and as I shall emphasize shortly, the Time 2 decision poses a different question just because it is not a case of first impression. None of these points turn on whether the Time 1 decision was correct or not, or on what the Time 2 decision should be. The point is that they are not the same decisions. Similar problems arise even when the earlier and later decisions involve the same level of government, on precisely the same constitutional issue. Suppose that at Time 1 the Court decided that capital punishment 118
119 120 121 122 123
124
This point does not hold for the polling interpretation. In that interpretation, voters are each expressing their own preferences, as discussed in Chapter 1. In the informational interpretation, however, voters are each guessing about a common, exogenously defined right answer. See Jeremy Waldron, Democratic Theory and the Public Interest: Condorcet and Rousseau Revisited, 83 Am. Pol. Sci. Rev. 1317, 1322 (1989). Thanks to Jack Goldsmith for emphasizing this point. Texas v. Johnson, 491 U.S. 397 (1989). Stenberg v. Carhart, 530 U.S. 914 (2000). Compare Johnson, 491 U.S. 397, with United States v. Eichman, 496 U.S. 310 (1990) (federal flag-burning statute); Stenberg, 530 U.S. 914, with Gonzales v. Carhart, 127 S. Ct. 1610 (2007) (federal partial-birth abortion statute). See Oliver Wendell Holmes, Law and the Court, in Collected Legal Papers 291, 295–96 (1920).
72
L L R
did not violate “evolving standards of decency.”125 Is this an informationally useful precedent if, at Time 2, the Court faces the same question again? Not necessarily, because standards may have evolved even further. In this example, the problem of change over time is right on the surface of the doctrinal test, but the same problem may arise in less transparent form. If at Time 1 the Court held that a certain rule of criminal procedure is or is not “implicit in ordered liberty,”126 does this mean that a Court facing the same issue at Time 2 is answering the same question? Even if, say, the two cases are a generation or more apart, and social conceptions of order and of liberty have changed in the meantime? Finally, and most generally, in a regime of precedent there will necessarily be a difference between the questions asked in cases of first impression and in all subsequent cases. At Time 1 the question is what legal rule the court should adopt, but at Time 2 the initial question is how much weight to give to the Time 1 precedent. Because the question at Time 1 always differs from the question at a later time, one cannot straightforwardly aggregate information across the divide between examination and re-examination of legal questions. Adherence to a regime of precedent itself undermines the commonality of questions that is necessary for the Condorcetian interpretation of precedent. One must not overstate these points, because a slight shift in the relevant questions does not deprive previous precedents of all informational value to current decisionmakers. Suppose that a decade ago, judges decided that a particular practice did not violate widespread standards of decency. Suppose also that judges today must decide whether the same is true today, under an evolving standard. It is surely relevant information to know that as of a decade ago, decency did not condemn the practice. But we cannot straightforwardly rely on the miracle of Condorcetian aggregation in a case like this, because we cannot lump together in a single notional voting group two different sets of judges deciding two different questions. The Condorcetian interpretation of Burke trades on a kind of metaphor, one not dissimilar to Burke’s own description of tradition as the “bank and capital of nations and of ages.”127 In the implicit metaphor of the common-law constitutionalists, the justices who have participated in a line of decisions are thought of as deciding in common, on the same question. But in many cases, no meeting of the many minds ever occurs.
125 126 127
Trop v. Dulles, 356 U.S. 86 (1958). Palko v. Connecticut, 302 U.S. 319, 325 (1937). Burke, supra note 94, at 183.
T C C L: I A
73
The metaphor here is affirmatively harmful. It obscures that the many minds participating in the line of precedent answer questions that shift subtly over time with shifting circumstances, changed legal contexts and novel applications of pre-established legal rules.
Independence and Herding Another critical condition for the Jury Theorem to apply is that the votes or guesses that are aggregated must be independent of each other.128 For brevity I will speak of the “requirement” of independence, but one must be careful. Nonindependence just reduces the number of effective votes or guesses in the pool. If there are ten voters, and nine utterly defer to a leader, guessing exactly as the leader guesses because the leader does so, then only one vote counts. This need not mean that the average competence of the group decreases. If the leader is of much higher average competence than the followers, it may even increase, at least if the follower would otherwise be more likely to guess wrong than right; Deference to opinion leaders can improve group performance.129 Nonetheless, under a broad range of conditions, the more independence, the more epistemic diversity goes into the group judgment, and the better the Condorcetian credentials of the precedent or tradition. This criterion is often applied too stringently. In one quite erroneous interpretation, independence is violated by mere deliberation, because people influence one anothers’ views. However, this is false. As discussed in Chapter 1, the independence required by the Jury Theorem is a statistical notion. Roughly, it means that Voter 2 does not decide to vote just as Voter 1 votes, however that is; it does not mean that Voter 1 cannot influence Voter 2’s independent judgment, through discussion or otherwise. Deliberation by itself does not compromise independence. That said, however, true failures of independence can arise from many causes. One cause is power—the power of ruling regimes, interest groups, or even court leaders. Suppose that the precedents of a given era were generated, not by statistically independent consideration by a panel of justices, but by a panel of justices who were political cronies following the lead of the White House, or the lead of outside groups, or who (like the
128 129
See Waldron, supra note 199, at 1326. See generally David Estlund, Opinion Leaders, Independence, and Condorcet’s Jury Theorem, 36 Theory & Decision 131 (1994).
74
L L R
justices of the Marshall Court) systematically deferred to the chief justice. The reduction of independence reduces the number of effective votes. When we turn to comparative institutional problems later in the discussion, the role of power will become important: arguments that impeach the Condorcetian credentials of other institutions, such as legislatures, on grounds of interest-group power or other process failures must consider the same possibilities with respect to judicial institutions. An important special case in which independence is violated involves informational cascades.130 Such cascades can arise even if all actors are rational and are attempting to make the best possible individual guess. Where each person has a small stock of private information, the rational guesser will do best to mimic the judgments of guessers earlier in the sequence because their total information swamps his own. The result can be a bad equilibrium in which a long sequence of people, deferring to erroneous guesses earlier in the sequence, make erroneous guesses themselves.131 This outcome is collectively bad but individually rational; each person followed the best strategy for themselves, but the outcome for all is a bad cascade. Cascades can occur in courts as well as in society generally; a line of precedents may represent little more than a rational decision by later judges to ignore their private information in favor of what earlier courts have said. Where this occurs, later decisions in the line of precedent are not independent contributions that add to the informational value of the whole; although decisions accumulate, the information implicit in the line of precedents does not increase. Although some argue on a priori grounds that precedential cascades are unlikely,132 they have in fact been documented under quite ordinary conditions.133 In some settings, cascades are fragile to small changes in information and motivations; although easily started, they are easily shattered as well.134 In markets, cascades can often be dispelled by releases of information, from government or firms. Cascades embody surprisingly little information; after the first few guessers, all others are rationally following the herd. 130 131 132 133 134
For an overview, see Sushil Bikhchandani et al., Learning from the Behavior of Others: Conformity, Fads, and Informational Cascades, 12 J. Econ Persp. 151 (1998). For a simple example, see Information Cascades and Rational Herding, http://www.infocascades.info/ (last visited Feb. 23, 2008). Eric Talley, Precedential Cascades: An Appraisal, 73 S. Cal. L. Rev. 87 (1999). Andrew F. Daughety and Jennifer F. Reinganum, Stampede to Judgment: Persuasive Influence and Herding Behavior by Courts, 1 Am. Law. & Econ. Rev. 158 (1999). See generally Christina Biccieri and Yoshitaka Fukui, The Great Illusion: Ignorance, Informational Cascades, and the Persistence of Unpopular Norms, 9 Bus. Ethics Q. 127 (1999).
T C C L: I A
75
Hence a small addition of information can cause actors later in the sequence to change their guesses, breaking the cascade.135 If this fragility to new information worked straightforwardly in the case of precedential cascades, one might have reason to think that most lines of precedent represent the informational contributions of many minds, rather than a cascade. However, precedential cascades are probably relatively robust. The reason is that “the passive nature of courts . . . means that erroneous decisions made by one level in a system of courts potentially remain uncorrected, even though all courts are trying to make principled judgments.”136 If the Supreme Court follows its usual practices of case selection, it will not grant certiorari in cases where all lower courts have cascaded to an agreement, and so will have no occasion to overturn the lower-court consensus. Litigants, seeing the lower-court consensus and knowing that Supreme Court review is unlikely, are unlikely to challenge the consensus in the first place. By virtue of the structure of the judicial system and the passivity of the courts, precedential cascades are plausibly less fragile than cascades in markets or society generally. Changes in motivation or incentives—in the reward structure of individual action—can also dispel cascades or prevent their arising in the first place. Where each person is attempting, not to make the best possible individual guess, but to make the guess that will prove most useful to the group, each guesser will ignore the earlier guesses and record her unaided view of the problem.137 When this happens, each guesser discloses more private information to be aggregated into the group judgment, and the quality of the group judgment improves. The same effect can arise not from incentives, but from cognitive quirks. Irrationally overconfident individuals contribute a great deal to the group; precisely because they are overconfident, they tend to contribute private information that can help to block or shatter cascades.138
The Burkean Paradox The significance of all this is to reveal a tension at the foundations of the Burkean view—what I will call the Burkean paradox. The paradox is that if many participants in the line of precedent or tradition followed the 135 136 137 138
Bikhchandani et al., supra note 130. Daughety and Reinganum, supra note 133, at 159. Cass R. Sunstein, Why Societies Need Dissent 66–67 (2003). See Antonio E. Bernardo and Ivo Welch, On the Evolution of Overconfidence and Entrepreneurs, 10 J. Econ. & Mgmt. Strategy 301 (2001).
76
L L R
precedent or tradition because doing so was a way to conserve on decisionmaking costs or improve their information, then the informational value of the line of precedent or tradition is lower to that extent; there are fewer independent minds contributing to the collective wisdom.139 The line of precedent or tradition itself becomes a kind of informational cascade, which lowers the total informational value of the whole. The sting in the problem is that a strategy that is individually rational for judges at any given time—following precedent—is harmful to all if followed by all, because it drains precedent or tradition of any epistemic value. Burkeanism might even be analogous to a kind of informational pollution, giving rise to severe problems of collective action: it is individually rational but collectively harmful for individual decisionmakers to rely on precedent or tradition, rather than their unaided reason. The Burkean paradox is not an all-or-nothing affair. Perhaps some judges in the stream of precedent or tradition have contributed independently, while some have not, in which case the informational value of that stream must still be discounted appropriately, but not to zero. Moreover, individual judges might adopt an intermediate approach, according to which they give some but not complete deference to the views of the past, and correlatively think for themselves to some degree or in some circumstances. This will alleviate the paradox, but only at the price of diluting the claimed benefits of following tradition in the first place. Burkeanism and the Burkean paradox have a sliding-scale relationship: the more strongly Burkean the judge’s decision rule, the stronger the paradox, while reducing the force of the paradox requires a reduction of the commitment to Burkeanism itself. In principle, judges might prevent the Burkean paradox from arising by issuing opinions that record their independent judgments while basing their actual decisions on precedent. In this model, later courts will have both the precedent-based decision, which is of low or nil informational value, and the earlier court’s independent judgment, which has positive value. Although this may sometimes happen, it is rarely observed in real-world institutions, and for good economic reason. Judges decide on the basis of precedent in part to conserve on the costs of decisionmaking for themselves. To incur the costs of forming an independent judgment that, by hypothesis, will not affect the current decision is to provide a
139
See Michael Moore, The Dead Hand of Constitutional Tradition, 19 Harv. J.L. & Pub. Pol’y 263, 269 (1996); cf. Eric A. Posner and Cass R. Sunstein, The Law of Other States, 59 Stan. L. Rev. 2006 (making a similar point in the setting of international and comparative law).
T C C L: I A
77
benefit only to future judges and is thus a contribution to an informational public good; some judges will free-ride, and the informational public good will still be underproduced. Empirically, later judges do not always have access to the reasoning of earlier judges, because of conclusory opinions, bargaining among judicial majorities to produce an appearance of consensus, and other phenomena that suppress information about earlier judges’ views. Consider that norms of apparent consensus on the Supreme Court restricted the number of concurring and dissenting opinions to low levels from 1800 until the early 1940s.140 The Burkean paradox bites hardest where precedent or tradition embody an information cascade, and, of course, much depends on the specifics of the cascade at issue. Some cascades can cause stampedes in the right direction, not the wrong direction (according to whatever exogenous definition of right answers we have assumed). And cascades are fragile in many ways, tending to shatter as new information is released or as circumstances change. But to the extent that precedent or tradition rests on cascades, rather than the independent and unaided contributions of many minds, Burkean praise for precedent or tradition is self-defeating. The best contributions to the stream of precedent or tradition are those in which individual judges, or small groups of judges, exercise their unaided reason. Those who rely on precedent or tradition on the ground that it is the “bank and capital of ages” make withdrawals from the common pool of information, for their private benefit; those who exercise their unaided reason contribute to the common pool, for the good of all future judges.
Random or Correlated Biases Another strand of Bentham’s critique of the common law centered on the professional self-dealing of “Judge and Company.” The judges and lawyers who administered the common law of England, on this view, constituted an elite guild that arranged the legal system in ways that benefited the legal profession, to the harm of the polity at large. Bentham interpreted this claim by diagnosing Judge and Company as promoters of their “sinister interests” (a motivational distortion) or as in the grip of “interest-begotten prejudice” (a cognitive distortion arising from a motivational distortion, akin to self-serving bias).141
140 141
See Lee Epstein, Jeffrey A. Segal, and Harold J. Spaeth, The Norm of Consensus on the U.S. Supreme Court, 45 Am. J. Pol. Sci. 362 (2001). Bentham, supra note 115, at 34–36.
78
L L R
However, we may also interpret Bentham’s core concern in less sinister terms, and in a way that fits comfortably within the framework of the Jury Theorem. In this interpretation, the decisions made by Judge and Company are less accurate than might otherwise be the case because a group of judges will suffer from positively correlated biases. Statistically, the degree of intragroup correlation of error is just a restatement of the independence condition, but putting it in these terms will bring out some central points. We must distinguish two issues, competence and the direction of error. Competence is the likelihood that an individual or an average group member will err. But if group members err, it is best if their errors are negatively correlated or at least uncorrelated—if they are as likely to err in one direction as another. Where error is randomly distributed, biased guesses in either direction wash out, on average, and true guesses prevail. As discussed in Chapter 1, the degree to which biases are correlated is a major determinant of group performance under the Jury Theorem.142 Where errors are highly correlated—where group members are subject to a common and systematic bias that causes them to err in only one direction—group performance can fall dramatically. The chance of a majority selecting the right answer varies inversely with the correlation of bias across the group. This implies that under plausible conditions a group of lower average competence, but with uncorrelated or negatively correlated biases, will outperform a group with higher average competence but positively correlated biases. From a Benthamite perspective, the problem is that Judge and Company are likely to have highly correlated biases, arising both from a common socioeconomic background and from their common professional training. Chapter 4 supplies detailed evidence for this claim. For now, suffice it to say that “most lawyers, whatever their background, have a narrow, professionally inflected perspective on governance.”143 To see the effect on decisionmaking, let us suppose, for example, that lawyers and hence judges are likely to systematically err on the high side when they attempt to estimate the costs and benefits of additional legal process in government decisionmaking, and thus will tend to require government to afford extra increments of process whose social costs are greater than its social benefits.144 As compared to a group whose errors are randomly
142 143 144
This is a major theme of Scott Page, The Difference (2007). Richard A. Posner, Law, Pragmatism, and Democracy 128 (2003). Cf. Jerold S. Auerbach, A Plague of Lawyers, Harper’s, Oct. 1976, at 414–12.
T C C L: I A
79
distributed, the guesses of a group of past lawyers—the judges who decided the cases now relied upon as precedent—are less likely to track right answers about the value of process. This reduces the informational value of the precedents to future judges who are trying to draw aid from many judicial minds in the past. The key point is not that judges are likely to get things wrong; it is that when they do get things wrong, they are likely to err in systematic rather than random ways. The correlation of biases across the decisionmaking group trades off against the competence of the group’s members. Even if judges are elite experts, of very high average competence, their likemindedness reduces group performance overall, narrowing the margin of superiority between the performance of a single judge or group of judges exercising their unaided reason today, and a judge or group of judges drawing on precedent. This is an updating of Bentham’s critique of legal and judicial professionalism: rather than point to motivational distortions on the point of lawyers and judges, one may also claim that common professional training reduces the informational quality of judicial decisionmaking, even if all judges and lawyers are sincere. In Chapter 4, I take this point further, arguing—on strictly epistemic grounds—that the Supreme Court should contain some nonlawyers.
Burke, Condorcet and Constitutional Law: Comparative Institutional Problems So far we have examined some problems that arise within the logic of the Condorcetian interpretation of Burke, both as applied to social traditions in general and as applied to judicial precedent in particular. Here I will turn more specifically to problems with constitutional precedent and tradition. In both cases, comparative institutional problems become critical. A judge or group of judges deciding constitutional cases at a given time may turn to the information embodied in precedent to improve their decisionmaking. The Condorcetian interpretation of Burke describes the conditions under which using precedent in this way will improve upon the unaided reason of current judges. Yet in the constitutional setting, there are always alternative sources of information available: the judgments of past constitutional framers, of current legislatures, and of the current executive. Even if the collective wisdom of many past judges is superior to that of one or a few present judges, on Condorcetian grounds, it does not at all follow that the collective wisdom of many past judges is superior to that of other institutions.
80
L L R
The same basic analysis applies to traditions, albeit with modifications. Social traditions stand on somewhat better epistemic ground than strictly judicial traditions; but where such traditions are refracted through judicial decisions, their Condorcetian credentials can be no better than that of the judges themselves, who must identify and apply the traditions. On the other side of the equation, legislators too can draw upon social traditions, and will have Condorcetian advantages in doing so. Ultimately, the same comparative institutional problems afflict common-law constitutionalism whether the focus is on precedent or tradition.
Past Judges and Framers In many cases, the principal target of common-law constitutionalists is originalism, which I will take to be the view that the Constitution should be interpreted according to its public meaning at the time of enactment. Which approach is better on informational grounds? Some originalists suggest that the highest source of latent wisdom is the text and original understanding.145 Shortly, I will argue that this claim is false in fact, but it is theoretically crucial. It emphasizes that common-law constitutionalists must consider the informational credentials of all relevant sources of law before declaring in favor of a regime of strong precedent. The testing case for both originalism and common-law constitutionalism occurs when there is a clear original understanding, a clear line of precedent, and the two are in conflict.146 Should the views of framers and ratifiers or the views of later judges prevail? Let us suppose that, despite the internal problems surveyed above, a stream of precedents does embody aggregated wisdom or information that is superior to that of a single judge or small group of judges today. Still, it is a very different question whether that stream of precedent is informationally superior to the views of framers and ratifiers in the past. Past judges have an informational advantage over framers and ratifiers of the even more remote past, all else equal. This is a central point of common-law constitutionalism: judges who lived after the founding era judges have had the opportunity to learn from experience147 or to acquire 145
146 147
Steven G. Calabresi, Text, Precedent, and the Constitution: Some Originalist and Normative Arguments for Overruling Planned Parenthood of Southeastern Pennsylvania v. Casey, 22 Const. Comment. 311 (2005). For putative examples, see Calabresi, supra note 145. Nothing in the discussion here turns on whether these examples are correct, a question on which I express no view. See Strauss, supra note 112.
T C C L: I A
81
new information about the costs and benefits of the framers’ choices. In Condorcetian terms, more information yields higher average competence. And fresher, more relevant information does as well: so many circumstances have changed so greatly, the founding era was so hopelessly unlike our own, that relying on the framers’ epistemic capacities is a recipe for disaster. While I believe that this point is a decisive objection to originalism, I note two possible originalist rejoinders. Some originalists believe that the framers were superior statesmen, possessing extremely high average competence. One might explain this putative higher wisdom of the framers either by a selection mechanism or by reference to the circumstances of the founding era. Perhaps the framers were the elite of the nation, selected from among the notables of their states, as opposed to later judges who often took the bench through cronyism or party service. Alternatively, the circumstances of constitution-making might have produced sharpened cognition in the framers, because the costs of making mistakes were higher, or because a time of crisis produced emotional states—of solidarity with contemporary citizens and with future generations—that induced more effort for the public good. Furthermore, one might claim that the framers acted behind a “veil of uncertainty”148 that produced a kind of impartiality: sincere framers attempting to guess the welfare-maximizing constitutional rule will know less than later generations, but will also be less biased because they are less likely to know the rule’s distributive consequences for themselves, their friends, family, profession, or social class. There is thus a tradeoff between the framers’ relative lack of information about the consequences of the rules they adopt, on the one hand, and their relative impartiality, on the other.149 In Jury Theorem terms, lack of information reduces competence, but pervasive uncertainty reduces the correlation of biases and pushes the whole group towards randomly distributed error, which is desirable.
The Epistemic Failings of Originalism Despite these points, I believe that originalism remains epistemically objectionable on Benthamite grounds. Originalism is the purest possible example in law of the “wisdom of the cradle” that Bentham mocked. The framer’s putative competence and impartiality were depreciating assets, 148 149
John C. Harsanyi, Morality and the Theory of Rational Behavior, in Utilitarianism and Beyond 39, 44–46 (Amartya Sen and Bernard Williams eds., 1982). Adrian Vermeule, Mechanisms of Democracy: Institutional Design Writ Small 31– 54 (2007).
82
L L R
and their epistemic value is now close to zero. The more circumstances have changed, the more likely that decisions should be made not by the framers of centuries ago, to whom our problems would have been unimaginable, but by decisionmakers with better current information, especially as the founding era and the last major episodes of formal constitutional revision (Reconstruction and the Progressive era) recede into the distant past. It is terminally absurd that the constitutionality of counterterrorism policies after 9/11, or of gun-control policies, is determined in part by poring over archival scraps and ambiguous statements from 1789 or 1791. However, none of this implies that judges are the current decisionmakers who should benefit from originalism’s epistemic bankruptcy. There are multiple alternatives to originalism, and judge-made constitutional precedent is just one. Below, I shall argue that current legislatures are the decisionmakers in the best position, insofar as epistemic considerations are concerned, to oversee common-law constitutionalism.
Past Judges and Current Legislatures Let us now turn to the most critical comparison: between (1) a line of judicial precedent that interprets a vague, general, or aspirational constitutional provision and (2) a statute that contradicts the judicial interpretation. In order to bias the inquiry in favor of common-law constitutionalism, I will assume that when the statute is enacted there is already a clear and consistent line of Supreme Court precedent to the contrary. This must be the best case for common-law constitutionalism, insofar as overriding the view of current legislatures is concerned. However, in cases of this type, the same Condorcetian arguments that common-law constitutionalists deploy against originalism also support judicial deference to the views of current legislatures. Precedent beats original understandings because the information of the constitutional framers is hopelessly outdated, but precedent might in turn be beaten by current statutes. If that is so, then the logical consequence of the informational interpretation of Burke is not robust judicial review, in the style of the Warren Court; rather, it is deference to legislative judgments. On this account, Burke leads via Condorcet to James Bradley Thayer, who argued for judicial deference to current legislatures in all but the clearest cases of legislative mistake.150 150
See James Bradley Thayer, The Origin and Scope of the American Doctrine of Constitutional Law, 7 Harv. L. Rev. 129 (1893).
T C C L: I A
83
To make progress on these issues, let us temporarily stipulate to some crucial assumptions that I shall examine in the subsequent discussion. First, suppose that the legislators who enact the statute and the courts who review it both address the common question of its constitutionality (whether or not the legislators also address its merits as policy). Second, suppose that the legislators who voted for a statute are on average more likely to get that question right than wrong, as are the judges who review it. Third, suppose that both legislators and judges sincerely vote their views of the statute’s constitutionality. Obviously all of these assumptions often fail to hold, but that is irrelevant. Legislators may not vote sincerely, but judges may not either, as a large body of recent literature shows.151 Theorists who justify precedent on jury-theoretic grounds typically assume sincere voting among judges, so we will do the same among legislators, in order to be able to compare apples with apples and thereby isolate the epistemic issues. Given these assumptions, several stylized facts about legislatures threaten to give them an overwhelming Condorcetian advantage: the sheer numerosity of their members, the diversity of their memberships, and their powerful institutional tools for acquiring information—including the relationship of representation between legislators and constituents, a major source of information that judges lack. I will take up each of these in turn.
Numbers and Selection Effects It is highly likely that legislatures with hundreds of members proceeding under simple majority rule will vote correctly, even if their average competence is near-random, just because of their sheer numbers. With a mere 301 voters—a smaller group than the House of Representatives—and an average competence of only .6, a majority will be almost certain to choose correctly between two options.152 In comparative terms, even if the average competence of Supreme Court justices is much higher than that of legislators, it is extremely unlikely that the nine justices sitting at any one time, or even the few dozens of justices who participate in a line of precedent, can do better than a highly numerous legislature, just because it is hard to improve on a vote that is almost certain to be correct. Suppose that the justices are much better lawyers than, say, the average senator or even 151 152
See generally Thomas J. Miles and Cass R. Sunstein, The New Legal Realism, U. Chi. L. Rev. (forthcoming), available at http://ssrn.com/abstract=1070283. See Christian List and Robert Goodin, Epistemic Democracy: Generalizing the Condorcet Jury Theorem, 9 J. Pol. Phil. 277, 287 tbl.1 (2001). If average competence in the group of 301 falls to.51, barely better than random, the majority’s chance of choosing correctly is about 64 percent. See id.
84
L L R
than the members of the judiciary committees. Not much follows. Numbers matter; a major point of the Jury Theorem is that many nearincompetents can do better than a small panel of elite experts. Quite obviously, the crucial question here is whether legislators’ average competence is greater or less than one-half (in the two-option case). I put aside the argument that legislators might always raise their competence to .5 by flipping a coin; this overlooks that an incompetent voter might not realize his own incompetence. The more serious problem is the type of selection effect discussed in Chapter 1: as Condorcet himself suggested, average legislative competence might slip below the crucial threshold of one-half precisely because of the large number of members. However, Condorcet was speaking of the unstructured legislatures of the eighteenth century, selected by and to some extent from a relatively uneducated population; it is unclear what he would make of a modern legislature, say Parliament or Congress. The representation ratio for Congress has fallen steadily over the course of American history; if the baseline is the average representation ratio across liberal democracies, then Congress should have more than 800 members, summing the House of Representatives and Senate together.153 Members of the current Congress are thus selected from a much larger pool than were members in 1789, and if elections have any tendency to select those of higher than average epistemic competence, the falling ratio should raise the average epistemic competence of the modern legislator. Better-than-random competence should thus be less of a scarce resource in the modern Congress, and if it is, then the selection mechanism is correspondingly less important. We need only enough confidence in the mechanisms of legislative selection and decisionmaking to believe that legislators are on average only slightly better to get the relevant question right than wrong (in the case of binary choices). If that undemanding threshold is met, the legislative advantage in numbers will tell.
Professional Diversity Legislatures are by and large more diverse than judiciaries, especially along professional lines. In the United States Congress, between 1960 and 2004, about 45 percent of the members were former lawyers;154 the rest 153 154
Emmanuelle Auriol and Robert J. Gary-Bobo, On the Optimal Number of Representatives, at 4 (July 2007), available at http://idei.fr/doc/wp/2007/optimal_number.pdf. See Chadd K. Kraus and Thomas A. Suarez, Is There a Doctor in the House? . . . or the Senate? Physicians in the U.S. Congress 1960–2004, Journal of the American Medical Association 292(17): 2125–29 (2004).
T C C L: I A
85
were former businesspeople, teachers, physicians, soldiers, and so forth. By contrast, all Supreme Court justices are lawyers, although no constitutional or statutory rule so requires. Professionals are inculcated with common skills and hence common prejudices; professional uniformity tends to create harmful groupthink, which we can interpret in Condorcetian terms as a high positive correlation of biases across the decisionmaking group. Bracketing the comparison of legislative and judicial competence, the greater professional diversity of legislatures produces greater cognitive diversity, and thus supplies a major epistemic advantage. Although this advantage must be traded off against a slightly lower chance of the legislative group getting right answers to highly technical legal questions, most constitutional questions that the Supreme Court decides are not highly technical; instead they are questions with large moral, factual, and causal components. As to questions of that sort, the cognitive diversity that accompanies professional diversity across the decisionmaking group is a decisive advantage of legislatures. As we will see in Chapter 4, adding decisionmakers who are worse than random—of lower competence than is a coin toss over two choices—can actually improve group performance by reducing the correlation of biases.
Legislative Organization, Representation, and Information Legislative institutions, especially the committee system and the “electoral connection” between legislators and constituents, bring legislators important information and thus help to push legislative competence up to the crucial threshold at which numbers tell. By virtue of the extreme internal specialization of the committee system, large legislative staffs, and the professionalism of the legislative career (at the federal level and in many states, though not all), modern legislatures are engines for generating and processing information. The seniority norm and the proliferation of subcommittees imply that a member of the modern House of Representatives spends many years becoming deeply expert in a narrow slice of public policy. 155 All this has two crosscutting effects. On the one hand, specialization increases the information and thus average competence of members, including members who have influence on constitutional questions, such as members of the judiciary committees. On the other hand, floor deference to committees and legislative leaders reduces the number of effectively independent votes; this is the problem of epistemic bottlenecks applied to the legislative process, as discussed in Chapter 1. The upshot is a smaller 155
Keith Krehbiel, Information and Legislative Organization (1991).
86
L L R
number of more competent votes, and the result is unclear. But it is not implausible to think that average competence in a modern legislature, highly professionalized and specialized by virtue of an elaborately reticulated committee system, is better than random on average. Even more important is that legislators are connected to constituents and thus have better information about the factual components and causal consequences of their constitutional decisions than do judges. We have seen, in Chapter 1, that Madison saw aggregate information as a major benefit of a numerous legislative assembly. The demands of re-election force legislators to leave the halls of government, to travel the land, and to meet constituents, including, now and again, ordinary people. Judges need never leave the cloister, or at least the bubble of professional prestige that surrounds them. There is, of course, a price to be paid for legislators’ superior information, in that the need to secure re-election can cause a distortion of legislators’ true views; although legislators are professionally diverse in their pre-political careers, they are all in a sense professional politicians, in modern national-level legislatures anyway. The main tradeoff, then, is between bias and information.156 Judicial procedures are designed to ensure equality of inputs, a form of evenhandedness. This is basically a leveling-down strategy of institutional design: by virtue of prohibitions on ex parte contacts and equal time for briefing and argument, no party has an advantage in presenting information to the judges. Moreover, the lack of any electoral connection on the part of judges gives them a remoteness from current politics and thus a kind of impartiality. The price of this evenhandedness and remoteness, however, is a relative dearth of facts and tacit knowledge. The very political insulation that pushes the judges’ competence up also tends to deprive them of information, which pushes it down. By contrast, legislative fact-finding uses a relatively unstructured hybrid process, part adversarial and part administrative or inquisitorial, which generates more information overall at the cost of greater bias or skew. Put differently, the task of generating information trades off against the epistemic efficiency of its use. Although political forces may cause legislators to use the information they possess in distorted ways, they may, by the same token, possess so much more information as to more than compensate for the epistemic inefficiency. Precisely because they are less insulated, legislators are more likely to possess information about changing 156
This paragraph draws on the illuminating discussion in Neil Komesar, Imperfect Alternatives: Choosing Institutions in Law, Economics, and Public Policy (1994).
T C C L: I A
87
political, social and economic conditions and changing public values, all of which are indispensable to common-law constitutionalism. The fairest conclusion is that legislators acquire more information, but process and use it in a more biased fashion at an individual level. Legislative diversity at the group level, however, helps to wash out the biases. What matters for Jury Theorem purposes is not bias in the abstract, but whether biases are correlated across the group. Because legislators as a group are more diverse than the judges, along professional and other margins, biases are less tightly correlated and will tend to cancel out; unbiased information will tend to rise to the top. Information and diversity work synergistically: legislative diversity helps superior legislative information, derived from the electoral connection and from legislatures’ open-ended fact-finding, to prevail at the group level even if individual legislators are more biased and less competent than individual judges. Although the electoral connection diminishes the competence of individual legislators, it is a fallacy of composition to assume that it therefore diminishes the competence of the legislature. It has been argued, convincingly, that all constituents will be biased on certain dimensions of policy—against free trade, for example—so that the biases of voters and representative legislators will also be highly correlated on those questions.157 This is doubtless sometimes true, but it is not plausibly true across all the varied domains of constitutional law, as to which there is a wide variety of public opinion. It is exceedingly unlikely that any official decisionmaking body will embody more cognitive diversity than the maximum of diversity to be found in the society; what matters is that however much or little diversity there is, representative and professionally heterogeneous legislators are more likely to capture and embody it than are narrowly professional lawyer-judges. Legislators hear from many more interests and social sectors, in both formal and informal ways. Aggregating across some dozens of judicial minds ameliorates this problem, but does not plausibly overcome the large advantages of numbers, professional diversity, intra-institutional specialization, and sheer information that legislators enjoy in the Condorcetian framework.
Endogenous Information and Rational Ignorance We must also consider the incentives of individual legislators and judges to acquire information, and the related problem of epistemic free-riding, 157
See Bryan Caplan, The Myth of the Rational Voter (2007).
88
L L R
an issue introduced in Chapter 1. The basic problem is that the amount of information each voter (whether legislator or judge) decides to acquire is endogenous to the size of the voting group. In large legislatures, each legislator has an incentive to remain rationally ignorant. The legislator will invest less in informing herself about the questions at hand if she estimates that, because of the legislature’s large size, she is extremely unlikely to be the decisive voter in any event. More generally, information and deliberation are collective goods in legislatures; each member may attempt to free-ride on the expertise of others. All else equal, these considerations tend to undercut the informational advantages of legislatures. On this view, the comparative advantage of courts is their relatively small size. By raising each judge’s probability of being the pivotal voter, the incentive to acquire information is greater, and it is easier for all members of the group to engage in mutual monitoring of free-riders. On a nine-member Supreme Court, each justice has a nontrivial chance of being the swing voter, and colleagues will notice and disapprove if a justice acquires no information before voting. However, these considerations are themselves in turn undercut by other features of legislatures. For one thing, legislators often abstain from voting, whereas justices very rarely do so. Rational abstention can allow uninformed legislators to, in effect, defer to colleagues who are known to be well-informed, thus raising the chance that the decisive vote will be cast by a well-informed voter.158 Of course legislators who abstain often do so, in fact, for less high-minded reasons, but again we are assuming sincere and public-spirited behavior on all sides in order to focus on the informational comparison between legislatures and courts. Furthermore, the relatively large size of legislatures merely reduces legislators’ incentives to acquire information as part of the legislative process. It does not remove whatever information they begin with or acquire for other reasons. Analogously, in electoral models of rational ignorance, it has been pointed out that “[i]t is reasonable to believe that voters are involuntarily exposed to a flow of political information in the course of everyday activities”; the consequence is that even under rational ignorance, “[s]uccessful information aggregation is possible because the information acquired by each voter goes to zero but it does so slowly enough to allow the effect of large numbers to kick in.”159 The endogeneity of information 158 159
Cf. Timothy J. Feddersen and Wolfgang Pesendorfer, The Swing Voters’ Curse, 86 Am. Econ. Rev. 408–24 (1996). Cesar Martinelli, Would Rational Voters Acquire Costly Information?, 129 J. Econ. Theory 225, 226–27 (2005).
T C C L: I A
89
plausibly fails to undercut the comparative epistemic advantage created by the sheer number of legislators. The difference between a Supreme Court of nine members and a Congress of more than four or five hundred is plausibly too large—the initial Condorcetian advantages of the latter group are plausibly too great—to be highly sensitive to differential incentives to acquire information. As discussed above, it seems plausible that legislators begin with much more relevant information than do judges. Legislators acquire information relevant to judgments of constitutionality—information about societal mores and evolving standards of decency, the consequences of constitutional decisions, and the facts on which constitutional decisions rest—just by getting elected in the first place, and through their close contacts with constituents, government policymakers and the broader society. By contrast, informational impoverishment is a principal cost of the judges’ insulation from the hurly-burly of government policy. So legislators’ relative disincentive to acquire more information through the legislative process, if it exists, may not matter very much in light of legislators’ initial informational advantages, just as a runner who gets a cramp may still win if her lead was large enough to begin with.
Legislatures, Precedent and the Burkean Paradox A final point about the court-legislature comparison: Legislatures occasionally draw upon their own precedents when fashioning new statutes. To the extent that legislatures draw upon their own precedents, they too are subject to the Burkean paradox. Does this undercut the comparative epistemic superiority of legislation? No; the point is glib. Precedent is not invoked as commonly in legislatures as in the courts, is not as formal when it is invoked, and has a different cost structure. Overall, legislation is not as resistant to change— “path-dependent”—as is the common law, as I will discuss in detail in Chapter 3. The key difference is that once legislatures have paid the fixed cost of enacting a new statute, they can simply implement their current views about what is best, no matter how great the departure from past statutes. By comparison, the greater the necessary adjustment, the more costly it is for judicial precedent to adapt and the more slowly judges will be able to do so. Finally, if judicial precedent makes judicial decisions better informed, then legislative precedent makes legislative decisions better informed. In either the judicial or legislative setting, there is a tradeoff: drawing upon
90
L L R
precedent exploits the information of the past, thereby improving decisions in the present, but adds nothing to the institution’s informational stock for future use—a stock that is continually depreciating as circumstances change. It is unclear whether, and under what conditions, legislatures or courts strike the optimal balance between the short-run benefits of exploiting extant information and the long-run benefits of generating new information. At a minimum, however, there is no basis for confidence that the judicial balance is systematically superior.
Courts and Legislatures: An Overall Comparison Considering the relevant variables together, I believe, shows the overall epistemic superiority of legislatures. Recall that the major determinants of epistemic performance, for groups, are numerosity, diversity and average competence. Legislators are far more numerous than judges, in the relevant cases; although modern national judiciaries are sometimes large, they bring only a small fraction of their memberships to bear on any given case. Legislatures are also far more diverse, which gives them clear epistemic advantages under the Condorcetian model. Pointing to this or that dim-witted or politically biased legislator, a favorite parlor sport among legal elites, overlooks that group performance can benefit from the presence of the dim-witted or politically biased if they diversify the group’s epistemic composition. The claim of superior judicial competence rests upon two pillars, freedom from politically-induced bias and the use of precedent. However, we have seen that both pillars are wobbly. As to the first, there is a chronic tradeoff between bias and information; the price of judicial impartiality is that judges are relatively uninformed, which itself reduces their competence. As to the second, the use of precedent by either legislatures or courts has ambiguous effects on institutional information. There is thus no clear superiority in overall judicial competence that offsets the inherent legislative advantages of numbers and diversity.
Past Judges and the Current Executive: Roosevelt and Condorcet Some of the same considerations hold, with appropriate modifications, for executive branch decisionmaking. In a sense the executive branch contains many more minds, at any given time, even than Congress, whose staff of twenty-odd thousand employees is dwarfed by the hundreds of thousands of (nonmilitary) personnel employed by the executive. This is
T C C L: I A
91
misleading because the internal organization of the executive branch is more hierarchical than that of legislatures, at least in American legislatures, with their relatively weak party discipline. In Condorcetian terms this hierarchy increases intragroup deference and reduces the effective independence of the aggregated “votes.” Nonetheless, the executive branch is a “they,” not an “it”; the “unitary executive” is a legal claim about for-mal lines of authority and removal that does not even purport to describe actual executive decisionmaking and has no direct relevance to a Condorcetian analysis. Because of the internal diversity of the executive branch and the multiplicity of its officials, it is an open question whether an executive decision aggregates more or less information than the decisions of judges in a regime of precedent. Executive branch policy analysts are, plausibly, highly competent in the sense that they are experts in particular fields. The harder question is how well executive expertise is actually used by decisionmakers higher in the chain; even if higher decisionmakers are sincere, and thus impose no political distortions on the analysts’ conclusions, they must in many cases select between or among competing analyses, and may do so well or poorly. Here too, however, examining only the current executive understates the case; executive offices routinely use precedents, in the sense of prior decisions that are informationally useful, whether or not legally binding. There is no basis for a systematic presumption that judicial precedent is informationally superior to executive decisions where these conflict. The executive’s epistemic competence, however great or small, is cumulative with the legislature’s in the cases I am discussing. My argument against epistemic legalism is restricted to the domain in which modern legislatures and executives work in tandem, rather than at cross-purposes, as where legislatures either legislate directly and entrust enforcement to the executive or where legislatures in effect delegate lawmaking power to the executive in the first place. This domain of legislative-executive cooperation, however, is extremely large; in the modern state, it covers most of the waterfront.
From Burke, Via Condorcet, to Thayer We have discussed the relative informational properties of common-law precedents and traditions, the original decisions of framers and ratifiers, the decisions of current legislatures, and of the current executive. I will try to draw together the discussion with some comparative statics. As to competence, the lower the average competence of past judges, and the higher the average competence of past framers or ratifiers, the less likely it
92
L L R
is that current judges should depart from or override clear constitutional text based on contrary subsequent precedent. However, where circumstances change over time, the informational advantage of later judges will become pronounced. Much depends on the rate of change in the legal and political environment, a point I amplify and underscore in Chapter 3. Yet when a constitution is very old, as in the American case, its informational content has typically depreciated nearly to zero. In America, originalism is a terrible epistemic bet. As the rate of change in the legal, political, social, and economic environment increases, however, legislation becomes the best epistemic repository of all. Here too some general comparative statics are possible. The lower the average competence of past judges, the higher the relative professional diversity of current legislatures, and the higher the average competence of current legislators, the less willing current judges should be to override current statutes on the basis of constitutional precedent. Even very competent judges today, possessed of precedents issued by very competent judges in the past, will rarely do well to overrule the decisions of current legislatures, especially because of legislatures’ diversity and sheer numerosity, both of which are important Condorcetian advantages. The very weapons that common-law constitutionalists rightly deploy against originalism can be used against them with even greater effect by Thayerians, because an epistemic analysis plays directly to the comparative advantage of legislatures. By invoking epistemic arguments against the originalists, the common-law constitutionalists tend to expose their flank. The first virtue of any theory of constitutional adjudication is to explain why courts should have the power to override legislative action. Originalism, for all its grievous epistemic problems, offers a straightforward account of that power: courts may do so when and because there is a validly enacted higher source of law authorizing or requiring them to do so. By contrast, the Condorcetian interpretation of common-law constitutionalism, and of Burke, leaves it not at all obvious why judges should rely on precedent or tradition to trump the views of current legislatures. The claim that many judicial minds are superior to one or a few does not give any leverage in making that additional, and more difficult, claim. To make the implications of these points more concrete, some observations about different legal and institutional contexts are useful. Consider the following: “Burkeanism as a sword” versus “Burkeanism as a shield.”160 An implication of what I have said is that there is no sound basis for using 160
Sunstein, supra note 96.
T C C L: I A
93
Burkeanism as a sword—meaning that there is no warrant for overriding the judgments of current legislatures or the executive by reference to precedent or judicial traditions. The most plausible consequence of the Condorcetian analysis is Thayerism. In this regime, legislatures might themselves be precedentialists and traditionalists, using the common law of past legislation, and broader social traditions, to inform new legislation. The overall regime would be one of Thayerism for courts, common-law constitutionalism for legislatures. Here we must be careful with the distinction between precedent (judicial traditions) and traditions more broadly. We have seen that, at the Supreme Court level especially, precedents may not embody the judgments of very many minds, at least compared to legislatures. Social traditions, on the other hand, may stand on a better footing, if they incorporate the judgments of a large number of people. However, there are two problems with any attempt to distinguish precedent from tradition as guides to judicial decisionmaking. One is the problem of epistemic bottlenecks introduced in Chapter 1. In the relationship between common law and custom there is the fundamental problem, identified by Bentham, that traditions are not self-defining or self-applying; if traditions are refracted through judicial decisions, then they have no better epistemic credentials than do the judges themselves, who may (inadvertently) distort or misconceive them in the process of identification and application. The second problem, relevant especially in constitutional cases, is that legislators too can rely on broader social traditions, not just on legislative or judicial precedent. If judges can discern and apply social traditions that have strong Condorcetian credentials, so can legislators; and to the extent there are right answers about what traditions permit or require, highly numerous legislatures have a built-in Condorcetian advantage over courts in discerning those answers. If Burkeanism as a sword is infirm, there is still a possible role for Burkean arguments that support legislative and executive judgments— Burkeanism as a shield. But this is not a very impressive role. Where constitutional texts are ambiguous, and precedent points in the same direction as enacted legislation or executive rules, judges will uphold what the legislature or executive has done. Here there is no obvious problem, and no real need to defend common-law constitutionalism as a distinctive approach to constitutional adjudication. Burkeanism as a sword is the theoretically crucial case for common-law constitutionalism, but it is also the very case where Condorcetian reasoning most strongly suggests that judges—either current judges or a notional judiciary aggregated over time—should defer to other institutions.
94
L L R
The Frankfurter canon. Justice Felix Frankfurter suggested that the decisions of past legislatures and presidents about how to allocate powers among themselves places a sort of common-law gloss on ambiguous constitutional texts, especially in areas of separation of powers and foreign affairs, where the constitutional texture is especially open.161 On the Condorcetian view, Frankfurter’s point is either important or banal, depending upon whether it is assumed that there is or is not a line of judicial precedent that construes the constitutional ambiguity in a way different than the other branches. If there is no such line of precedent, then on Condorcetian premises it is banal that current judges should rely on the implicit collective judgment of legislators and presidents over time. Those judgments will be made by a statistical group far more numerous, and at least as expert, as the small group of current judges. If there is a line of contrary precedent, however, Frankfurter’s idea is more cutting. It suggests that the aggregated wisdom of many judges, over time, is still inferior to the aggregated wisdom of legislators and executive officials, over time. In this latter case, Frankfurter’s view is a special case, in the area of separation of powers, of the claim that Burkean common-law constitutionalism should not be used as a sword—at least not when Burkeanism is interpreted in informational terms. State legislatures and re-enactments. Here is another special case of the same point. Suppose that at Time 1, Supreme Court justices issue a precedent decision invalidating state laws. At Time 2, many states re-enact those laws. Should the Court acquiesce at Time 3, or enforce its contrary precedent? In 1972, for example, the Court invalidated most extant deathpenalty schemes; but after a wave of state re-enactments, the Court changed course and approved the death penalty (with constraints) in 1976.162 On Condorcetian grounds, the answer is (almost certainly) that the Court should acquiesce. Assume, however heroically, that state legislatures and the Court are addressing the same question at all times—whether a given state statutory rule complies with the federal constitution—and that there is a right answer to this question. The aggregated current judgments of a massive number of state legislatures are superior, in terms of the Jury Theorem, to those of the group comprising the judges and justices who sat in the precedent case(s). It is irrelevant that the legislators’ judgments are themselves sorted into groups (state legislatures) who make separate 161 162
Youngstown Sheet and Tube Co. v. Sawyer, 343 U.S. 579 (1952) (Frankfurter, J., concurring). Compare Furman v. Georgia, 408 U.S. 238 (1972) (invalidating most extant death penalty schemes), with Gregg v. Georgia, 428 U.S. 153 (1976) (validating death penalty after state reenactments).
T C C L: I A
95
collective decisions under internal majority rule. The Jury Theorem applies even to the purely notional and statistical group comprising all the state legislators taken together, so long as the Theorem’s other conditions are met. In general, Condorcetian reasoning does not supply a convincing interpretation of common-law constitutionalism, particularly the latter’s explicitly Burkean claim that the limits of human reason support reliance on constitutional precedent and tradition. The internal problem is simply that the Jury Theorem’s conditions will, quite often, not apply to any realistic description of the process of common-law constitutionalism. The comparative institutional problem is that Condorcetian reasoning itself defangs common-law constitutionalism in the most theoretically crucial cases—where constitutional precedent is used to trump clear constitutional text and history, or to trump the judgments of current legislatures and executive officials.
Cumulative Epistemic Competence? Crucial cases of that sort also point up a major flaw in the cumulative argument for epistemic legalism. The argument, recall, is that even if legislatures are epistemically superior to courts, the addition of a layer of judicial review may still improve the overall epistemic capacities of the system. But this fails in the core cases to which common-law constitutionalism speaks, where the question is whether the courts should prefer their own views, based on precedents, traditions, customs, and their own epistemic competence, to that of legislators. Here there is no accumulation of epistemic competence but instead a confrontation of two institutions’ opposing positions. If the judges prefer their own view, and invalidate the statute, they are not adding epistemic competence to the system but subtracting legislators’ epistemic competence from the rulemaking process. The resulting rules are made by judges alone. The rules may or may not be epistemically superior to ones made by legislators alone—that is a separate question, to which the bulk of this chapter has been devoted—but their embodied epistemic competence just is what it is; it does not arise through the accumulation of competence across institutions.
Thayerism, Cumulative Review, and Epistemic Free-Riding Furthermore, Thayerism reduces a problem of epistemic free-riding that can arise under common-law constitutionalism. Thayer famously worried
96
L L R
that judicial review would dilute legislators’ sense of constitutional responsibility.163 The concern can be interpreted in purely nonepistemic terms, as a concern about moral hazard; but I want to offer an epistemic interpretation. Legislators who anticipate constitutional judicial review may rationally invest less in gathering and processing information, for the reasons discussed in Chapter 1. Precisely because they know that judges will be trying to catch their mistakes, they may commit more mistakes. The free-riding might even be mutual, in a case of “after you, Alphonse.” Legislators and judges might, on this picture, both reach fewer right answers than would either institution acting alone. This point by itself does not show that courts should defer to legislators rather than vice versa. What it does show is that adding more minds does not necessarily improve matters, also as argued in Chapter 1. If legislatures are epistemically superior to courts, in the domains we are discussing, then courts should defer to them precisely in order to reduce the pernicious epistemic incentives. The epistemically superior institution is not helped, but hurt, by adding layers of review conducted by epistemically inferior officials. The cumulative argument for epistemic legalism— that even if judges are epistemically inferior, adding judicial review can improve the system overall—fails if and to the extent that these perverse effects obtain. To be sure, this epistemic interpretation of Thayer’s concern does not show that legislatures indeed are epistemically superior; that must be separately established, and I have tried to do so. But it does show that if legislatures are epistemically superior, a fallback argument for epistemic legalism as a cumulative improvement of the system does not succeed.
Conclusion: Common-Law Legislatures, Thayerian Courts My major conclusion is simple. The limits of reason, interpreted in Condorcetian terms, do not support a preference for common-law constitutionalism. Rather, if the goal is to maximize the epistemic quality of constitutional lawmaking, the most plausible legal regime would be one of legislative constitutionalism and Thayerian, highly deferential courts. In this regime, common-law constitutionalism is entrusted to legislatures, who resolve ambiguities and update constitutional values over time by enacting statutes.
163
See James Bradley Thayer, John Marshall 107 (1901).
Chapter 3
The Constitutional Common Law: Evolution
I
now turn from an informational interpretation of common-law constitutionalism to an evolutionary one.164 The Condorcetian interpretation of Burke assumes the existence of right answers (although not necessarily preference-independent ones) in constitutional cases; posits that judges are sincerely attempting to find those answers; and posits that judges are more likely to find those answers by relying on common-law precedents than on their “private stock of reason.” In the evolutionary interpretation, by contrast, no judge need be seeking to produce right answers. Different judges have different motivations, ideologies and biases. Nonetheless, the claim runs, some mechanism causes the uncoordinated action of biased judges to produce a body of common law that is epistemically impressive, containing latent collective wisdom. As discussed in Chapter 1, if such a mechanism exists, it would be a species of invisible-hand explanation, in which collective wisdom would arise from human action but not from human design.
164
For the evolutionary interpretation of Burke, see, e.g., David A. Strauss, Common Law Constitutional Interpretation, 63 U. Chi. L. Rev. 877, 879 (1996); Cass R. Sunstein, Burkean Minimalism, 105 Mich. L. Rev. 353, 368 (2006); Ernest Young, Rediscovering Conservatism: Burkean Political Theory and Constitutional Interpretation, 72 N.C. L. Rev. 619, 653–56 (1994).
97
98
L L R
I begin with some essential preliminary work that distinguishes evolution to efficiency from evolution to epistemic accuracy, and that also shows how these two ideas can and have been connected to each other in the work of Hayek and modern scholars of law and economics. I then turn to the evolutionary interpretation of Burke, and finally examine recent economic work that models the common law’s possible evolution towards full informational efficiency. Throughout, I pursue the usual two-pronged approach: I question the internal logic of evolutionary accounts of the ordinary common law, and then argue that those accounts, even if correct on their own terms, cannot successfully be transposed to the setting of the constitutional common law.
Efficiency and the Limits of Reason We will see, throughout the chapter, that legal theorists, economists and philosophers connect evolutionary models of law with efficiency, somehow defined. (Of course there are several different conceptions of efficiency; I will try to make clear, as the discussion proceeds, which particular sense is at issue in any given setting). In biology, evolution through natural selection and other mechanisms is thought to promote efficient adaptation of organisms to local environments; here, to say that an organism is well-adapted is to say that it operates efficiently in an engineering sense. How then does efficiency relate to the limits of reason, to Hayek, and to Burke? In modern law and economics, as we shall see, a key claim is that common-law institutions evolve towards efficiency. Is this an epistemic claim? Or simply a claim that the common law minimizes deadweight losses, and thus maximizes social wealth? To understand the issues, it is best to begin with Hayek’s evolutionary account of the common law.
Hayek and the Common Law For legal theorists, the most obviously relevant of Hayek’s many analogies to the market—morality is a spontaneous order like the market, custom is a spontaneous order like the market, and so on—is the Hayekian claim that the common law is a spontaneous order that incorporates dispersed information, and thus proves superior in some sense to the collectivist and intentional legislation that a Benthamite might favor.165 Here again, 165
See generally Friedrich A. Hayek, Law, Legislation, and Liberty, Vol. I: Rules and Order (1973).
T C C L: E
99
Hayek or at least the Hayekian tradition conflates several different things that must be disentangled. Moreover, the version of Hayek’s argument on which he finally settled is far narrower than usually assumed, and has few concrete implications. The first step is to distinguish Hayek’s ideas from several near and not-so-near relatives. To begin with, we must distinguish (1) an intentionalist argument for the beneficial properties of the common law from (2) an invisible-hand argument for the beneficial properties of the common law. As the common law is a diachronic institution, extended over time, invisible-hand arguments in this domain are inevitably of the evolutionary variety. Moreover, we must distinguish (2a) evolutionary arguments at the level of particular common law rules from (2b) evolutionary arguments at the level of the whole society, including the legal system. It turns out that Hayek’s view is (2b), not (1) or (2a), and this blocks an easy application of Hayek’s views to the particular rules of the legal system. Position (1) can be associated with Judge Richard Posner,166 while (2a) later became a standard view in law-and-economics.167 Position (1) is simply the argument that judges aim (or, in a normative variant, should aim) to maximize social wealth. That position is distinctly un-Hayekian, because it sees judges as a cadre of centralized, even Benthamite decisionmakers who rely for the most part upon their first-order reason to develop efficient rules, subject only to the weak constraints that higher-level legal rules must be respected and that precedents might convey useful information. Thus true-blue Hayekians reject the Posnerian vision of efficiencyseeking judges (and are especially suspicious of the Supreme Court, which they see as a centralized rulemaker). These Hayekians see custom as a genuine and highly beneficial spontaneous order, but deny that the common law system, at least as it currently operates, actually incorporates custom.168 Position (2a) is the argument that the common law will, at the level of individual rules, evolve towards efficiency. The basic mechanism is that inefficient rules will impose deadweight losses and will thus be more likely to be challenged by litigants. The losers from the rule will be willing to pay 166 167
168
Richard A. Posner, Economic Analysis of Law 321–27 (1973). See Paul H. Rubin, Why is the Common Law Efficient?, 6 J. Legal Studies 51 (1977); George L. Priest, The Common Law Process and the Selection of Efficient Rules, 6 J. Legal Studies 65 (1977). John Hasnas, Hayek, The Common Law, and Fluid Drive, 1 N.Y.U. J.L. & Liberty 79, 92–98 (2005); Todd J. Zywicki, The Rise and Fall of Efficiency in the Common Law: A Supply-Side Analysis, 97 Nw. U. L. Rev. 1551, 1628–29 (2003). However, Hayekians of this stripe think that the common law was efficient in an earlier period, when it competed with other law-supplying institutions, such as equity and church law. See id. at 1620.
100
L L R
more to overturn the rules than the gainers will be willing to spend to defend them (this follows from the assumption that the rule creates a deadweight loss) and so, over time, inefficient rules will tend to be eliminated. Among the many implicit conditions that make this hypothesis fragile and somewhat lacking in generality, I will confine myself to one observation, to be expanded below. Even if all the conditions required by this mechanism do actually obtain, it only shows that the common law will evolve towards efficiency. By itself, it says nothing about the rate at which it will do so. The crucial question then becomes the rate at which the surrounding environment changes.169 If it changes slowly, then it is plausible that the common law will evolve to become quite efficient in the static or slowly changing environment. If, however, the nonlegal environment changes rapidly, then the common law will constantly be facing a moving target and may be much less efficient at any given time than, say, statutes intentionally designed by Benthamite legislators, or even statutes designed by ignorant or self-interested legislators. 170 In a rapidly changing environment the common law may be constantly be evolving from 0 percent efficiency to 5 percent efficiency, over and over again, and then statutes need not be very efficient at all to be comparatively superior. Hayek, however, does not offer evolutionary arguments at the level of individual rules in any event. Rather his arguments operate at the level of the whole society, including the legal system.171 The suggestion is that the common law and related liberal institutions, such as the market, enable the social coordination of individual expectations, and that a society with a legal and economic system that coordinates expectations in this way will outperform competitors.172 This argument is ill-defined. “Hayek is noticeably unspecific about the precise mechanism by which selection works. On different occasions, there are hints of different models of selection, based variously on imitation, inter-group migration, competition for resources, differential rates of reproduction, and wars of extermination; but none is developed in
169 170
171 172
Cf. Jon Elster, Introduction, in Rational Choice, at 26 (Jon Elster ed., 1986). Cf. Don Herzog, Poisoning the Minds of the Lower Orders 22–23 (1998) (“the traditionalist has to believe that the pace of social change is restrained, so that we have time to tinker with our policies and that tinkering is good enough”). See Gerald F. Gaus, Hayek on the Evolution of Society and Mind, in The Cambridge Companion to Hayek 232, 238 (Edward Feser ed., 2006). Todd J. Zywicki and Anthony B. Sanders, Posner, Hayek, and the Economic Analysis of Law, 93 Iowa L. Rev. (forthcoming 2008), available at http://ssrn.com/abstract=957177.
T C C L: E
101
any detail.”173 There is a burgeoning literature on “legal origins” suggesting that nations whose legal system descend from the common law outperform nations with other types of legal system, on various measures of economic and political well-being, due to the security of economic expectations that a common law system provides.174 This literature might suggest that the broad thrust of Hayek’s view was correct, even if he did not understand the mechanisms that make it so. Yet in the Conclusion to the whole book, I will suggest that the legal origins literature has recently shifted its thesis, becoming now a claim that market-supporting strategies of regulation which can stem as much from statutes and civil law as from common law—promote economic development. On this version, the legal origins literature does not even provide an indirect defense of the common law per se. However, I want to focus on a different problem with the systemic version of Hayek’s argument for the superiority of the common law. Even if well-specified, convincing, and empirically validated, the argument has few implications for action within an ongoing common law system, for the following reasons. The systemic version of the Hayekian approach is extremely narrow: it underwrites no conclusions whatsoever about the efficiency or desirability of any particular common law rule. That the common law as a system is superior to (say) a civil law system, as measured by some criterion of fitness, just means that one package of institutions and rules is superior to another package of institutions and rules. Any individual rule in the superior system might be inferior to the alternative rule found in the other system. Moreover, the general theory of second best175 implies that the superior system might actually contain more rules that are inefficient or unfit than the inferior system. If a given system departs from the optimum in one respect, as it inevitably will, then the theory of second best suggests that other departures from the optimum are necessary. The system whose rules most closely approach the optimum, evaluating the efficiency of rules one by one, might well be the inferior system. It is even possible that all rules in a given system are suboptimal and indeed inferior to those of 173 174
175
Robert Sugden, Spontaneous Order, in The New Palgrave Dictionary of Economics and the Law 485, 494 (Peter Newman ed., 1998). Rafael La Porta, Florencio Lopez-de-Silanes, and Andrei Shleifer, The Economic Consequences of Legal Origins, 46 J. Econ. Lit. (forthcoming 2008), available at http://www.economics. harvard.edu/faculty/shleifer/files/jel.nov_combine.pdf.pdf; Paul G. Mahoney, The Common Law and Economic Growth: Hayek Might be Right, 30 J. Legal Stud. 503 (2001). R.G. Lipsey and Kelvin Lancaster, The General Theory of Second Best, 24 Rev. Econ. Stud. 11 (1956).
102
L L R
the competing system, but that because of the particular structure of their interaction, the given system is superior overall. Because systems compete as packages, an evolutionary account of the superiority of the common law pitched at the system level tells us little about particular questions within an ongoing common law order.
Hayek, Common Law, and Efficiency We are now in a position to understand that Hayek’s account of spontaneous order in general, and hence of the common law in particular, is ambiguous or ambivalent on the question of efficiency. In his central work, on the division of knowledge, Hayek’s argument for the price system as against central planning combined efficiency notions with the limits of reason in an illuminating way. Hayek’s central claims, in this arena, are that the price system makes the best possible use of dispersed information and enables localized and tacit knowledge to be used to social advantage. On this account, the market is itself a form of social planning, indeed a highly efficient form; it is just not a system of central planning. In other writings, Hayek advanced a consequentialist, but nonutilitarian, account in which the division of knowledge promoted “social coordination” and the pursuit of individual ends, but not efficiency as such.176 Finally, in work towards the end of his career, Hayek returned to efficiency-based claims, this time at the level of social selection; he suggested that social systems of rules—emphatically including legal rules— competed with the systems of other societies, and that the more efficient would prevail.177 Despite the ambiguities in his thought, one of Hayek’s major contributions, and one of the reasons for his enduring appeal to many, is the links he forges between two major families of ideas: information and epistemic capacity, on the one hand, and evolution through selection, on the other.
Efficiency and Information in Law-and-Economics Hayek was a nonstandard economist; what about the economic mainstream? In law and economics since about 1960, the main efficiency concept is some version of Kaldor-Hicks efficiency or wealth maximization, 176 177
Eric Mack, Hayek on Justice and the Order of Actions, in The Cambridge Companion to Hayek 259 (Edward Feser ed., 2006). Celia Lessa Kerstenetzky, Hayek and Popper on Ignorance and Intervention, 3 J. Institutional Econ. 33, 38–39 (2007).
T C C L: E
103
in which the evaluative criterion is whether legal rules and institutions maximize aggregate social wealth, avoiding deadweight losses. There are many well-known conceptual issues here, about whether wealth is a value, whether it is a useful proxy for values such as welfare or utility, and so forth; although I touch briefly on those issues below, they are not essential to my points. All I mean to suggest here is that the law and economics account of the common law’s efficient evolution itself has indirect epistemic significance. The key point is that the invisible-hand processes that are said to produce efficient common law, even if judges do not intend to create efficient legal rules, substitute for the informational deficits that would plague efficiency-seeking judges. As I mentioned, Judge Richard Posner’s original argument for the efficiency of the common law was that common-law judges intentionally maximized efficiency, in part because their inability to effect significant redistribution left them no other option.178 This claim was subjected to withering fire, with one of the major criticisms being that common-law judges lack sufficient information to select efficient rules. Here was a major impetus for the development of invisible-hand models, which are by their nature less information-demanding; in these models, individual judges do not need either to attempt to promote efficiency, or to acquire sufficient information to do so. This point, about evolution as an epistemic substitute for the intentional pursuit of efficiency, was made explicit in one of the seminal papers of this branch of law and economics, which argued that “the efficient rule situation noted by Posner is due to an evolutionary mechanism whose direction proceeds from the utility maximizing decisions of disputants rather than from the wisdom of judges.”179 In more recent law and economics work, the connections between information and efficiency become even more explicit. The first generation of law and economics invisible-hand models were demand-side invisible-hand models; evolution to efficiency is driven by litigants’ demand for efficient law, expressed through decisions whether to settle or litigate cases, rather than by judicial behavior. Important recent work by economists Nicola Gennaioli and Andrei Shleifer, however, offers supply-side invisible-hand models of the common law’s efficiency. In these models, judges are not intentionally promoting efficiency, which still arises (if it does) only as an unintended byproduct of judicial action. However, judges attempting to promote their own policy preferences 178 179
Posner, supra note 166. Rubin, supra note 167, at 51.
104
L L R
unintentionally contribute to the informational efficiency of law by offering material distinctions of earlier cases, thereby adding information content to the legal rules. Here efficiency is interpreted in a special sense, to mean that the common law evolves, through the process of distinguishing cases, to incorporate all material information. Gennaioli’s and Shleifer’s work can thus be understood as offering something like an epistemic account of common law efficiency. Gennaioli and Shleifer trace this idea to an insight offered by Judge Benjamin Cardozo, to the effect that “[t]he eccentricities of judges balance one another . . . out of the attrition of diverse minds there is beaten something which has a constancy and uniformity and average value greater than its component elements.”180 I examine these views in detail later in the chapter.
Burke, Evolution, and Efficiency How does all this relate to Burke, and to the constitutional common law? The short answer is that I will use these models to give content to the vague gestures towards the test of time, or evolution, present both in Burke and (especially) in neo-Burkean theorists of common-law constitutionalism. Burke’s account of tradition, custom and the common law, outlined in Chapter 2, is evocative but unclear. Burke must be interpreted, or construed, to have any utility or to hold any lessons for legal and constitutional theory. We have seen that one standard interpretation of Burke is evolutionary, but this idea is opaque in its turn; one must draw upon wellspecified evolutionary models to give content to the standard evolutionary interpretation of Burke. I will therefore read Burke in light of the standard demand-side law and economics models of common law efficiency, and then turn, later in the chapter, to the more recent supply-side models of informational efficiency. The latter, in particular, straddle the purely informational and Condorcetian account of Burke I examined in Chapter 2, and the evolutionary account I examine here. The main conclusions I will draw is that the limits of reason lead, in legal theory anyway, to Bentham and Thayer, not to Burke. In Chapter 2, the attempt to interpret Burke by way of Condorcet led to James Bradley Thayer’s idea that courts should systematically defer to legislatures in constitutional matters, or so I suggested. Likewise, the attempt to interpret
180
Nicola Gennaioli and Andrei Shleifer, The Evolution of Common Law, 115 J. Pol. Econ. 43, 44–45 (2007) (quoting Benjamin N. Cardozo, The Nature of the Judicial Process 177 (1921)).
T C C L: E
105
Burke via Darwin and Hayek leads to Thayer, and to a broadly Benthamite or neo-Benthamite skepticism about the constitutional common law.
The Mechanisms of Legal Evolution Exactly what evolutionary mechanism, or mechanisms, might underpin the common law? An evolutionary selection mechanism requires variation in the relevant population (of organisms or legal rules); heritability, so that favored variations can be maintained in the population when they arise; and differential reproductive advantage (or “fitness”), meaning that favored variations are likely to replace competitors.181 In evolutionary analyses of the common law, the first requirement is typically said or assumed to be satisfied by variation in the biases or preferences of judges or litigants, while the second is satisfied by the force of precedent. However, the third requirement is the most problematic; here the analogy to Darwinian processes becomes particularly obscure. What mechanism corresponds to natural selection of relatively fitter organisms? The most familiar attempt to answer this question involves the demand-side models of common-law evolution that have been prominent in law and economics. Here “demand-side” means that the common law’s claimed evolution towards efficiency is powered by the selection of cases for litigation—by litigants’ decisions about what cases to appeal or settle.182 I will examine both internal and institutional problems with the demand-side models. First, as an internal matter, these models suffer from a kind of fragility; they are not robust against slight changes in the assumptions. Second, in comparative institutional terms, there is no straightforward way to transpose these models to the setting of constitutional common law. Even if constitutional common law tends to evolve towards efficiency, and even if it is desirable that constitutional be efficient, legislation may also tend towards efficiency, and for the same reasons. Moreover, in a rapidly changing environment, legislation may at any given moment be more efficient than constitutional common law, even if the latter systematically tends towards efficiency while the former does not.
181 182
Stephen Jay Gould, The Structure of Evolutionary Theory (2002). For supply-side analysis, based on competition among institutions providing legal rules, see Zywicki, supra note 168. Zywicki also argues that an increase in the strength of precedent in the nineteenth century exacerbated various problems with the common law, such as rentseeking. See id. at 1565.
106
L L R
Demand-side Models: Internal Problems I will not attempt a comprehensive overview of this massive literature.183 Instead I will single out two relevant problems with the demand-side models: the fragility of their assumptions, and their silence about the rate at which common-law precedents converge to efficiency. The former problem is better known, but the latter is even more serious, because it suggests that in rapidly changing environments the tendency of common law to converge to efficiency—even assuming it exists—will not matter very much; in particular, it will provide no comparative advantage over legislation.
Fragility The basic engine of the demand-side models is selection pressure by litigants: if inefficient precedents are challenged more frequently, while efficient precedents are more often left in place, the law will tend to efficiency over time. The basic aspiration of the literature is to propose mechanisms that will show why, and under what conditions, litigants will tend to differentially litigate inefficient precedents. In the model that inspired much of the work in this area, the basic idea is that inefficient precedents inflict deadweight losses; the losers will thus have an incentive to pay more to attack the precedents than their beneficiaries will pay to defend them.184 Many second-decimal refinements have been proposed, but the underlying ideas are similar, so I will call this the “basic model” and take it as the subject of discussion. The basic model has been subjected to several major lines of critique, within a broadly rational-choice framework.185 First, the basic model assumes that precedents can only be reaffirmed or overruled. If they can also be further entrenched, however, then differential litigation might actually tighten the grip of inefficient precedents on the law.186 Second, the basic model assumes that all classes of litigants have, on average, equal
183
184 185
186
For a recent summary, see Paul H. Rubin, Why Was the Common Law Efficient?, in The Origins of Law and Economics: Essays by the Founding 383–395 (Francesco Parisi and Charles Rowley eds., 2005). Rubin, supra note 167; see also Priest, supra note 167. For a critique drawing on behavioral economics, see Adam J. Hirsch, Evolutionary Theories of Common Law Efficiency: Reasons for (Cognitive) Skepticism, 32 Fla. St. U. L. Rev. 425 (2005). Cf. Gillian K. Hadfield, Bias in the Evolution of Legal Rules, 80 Geo. L.J. 583 (1992). I will bracket such issues in the discussion here. Richard A. Posner and William M. Landes, Legal Precedent: A Theoretical and Empirical Analysis, 19 J.L. & Econ. 249, 272 (1976).
T C C L: E
107
stakes and an equal interest in the future consequences of precedent. But if different classes of litigants are differentially organized, perhaps because they have systematically different stakes or because some are repeat players while others are single-shot litigants, then precedent can just as well evolve in inefficient directions.187 There may be eventual convergence, but to rules favoring organized groups, not rules favoring efficiency. In general, rent-seeking litigation is as much a problem for models of common-law efficiency as rent-seeking lobbying is for models of legislative efficiency,188 a point I will return to shortly. According to one historically-inflected version of this hypothesis, common law was efficient in the eighteenth and nineteenth centuries, but is no longer so, because of the large-scale organization of rent-seeking litigation groups189 or (in a supplyside version) because of the decline in competition by different court systems.190 The basic model can and has been refined to try to account for these critiques, but only by making more stringent the conditions under which precedent converges to efficiency.191 After some two generations of discussion, the fairest assessment is that the thesis of common-law efficiency (at least as driven by demandside effects) is a possibility theorem and no more. Under imaginable conditions the common law will systematically tend to converge towards efficiency. However, slight changes in the initial assumptions seem to produce large effects, driving the implications of the models hither and yon and leaving the basic thesis unresolved. The natural response is to call for some facts; there is a related empirical literature on “legal origins,” or the economic effects of common-law legal systems versus those of civil-law legal systems, which I will discuss in the conclusion to the whole book. In the abstract, however, it can at least be said that the mechanisms in view are insufficiently robust to inspire any confidence.
Environmental Change Suppose the basic demand-side mechanism is not fragile, but robust, and ignore comparative institutional questions, which I shall take up shortly. What implications does the basic model have? Surprisingly few, for a simple reason that has not been much discussed in the literature. 187 188 189 190 191
The authors of the basic model recognized this point. See Rubin, supra note 183. See Einer L. Elhauge, Does Interest Group Theory Justify More Intrusive Judicial Review?, 101 Yale L.J. 31 (1991); Gordon Tullock, The Case Against the Common Law (1997). Paul H. Rubin, Common Law and Statute Law, 11 J. Legal Stud. 205 (1982); Tullock, id. See Zywicki, supra note 168. See, e.g., Keith N. Hylton, Information, Litigation, and Common Law Evolution, 8 Am. L. & Econ. Rev. 33 (2006).
108
L L R
Efficiency is always relative to the constraints established by the environment. In the economic sense, in which all economic factors are put to their highest-value use and all Pareto-improving trades are consummated, these constraints arise from available resources and technology (parametric constraints) and from competition by other economic actors (strategic constraints). In the biological sense, the optimal adaptation of organisms is determined by natural resources and competition from other organisms. In either case, the claim that some population of agents—organisms or firms or legal rules or whatnot—evolves towards efficiency is importantly different than a claim that the population reaches efficiency. A crucial variable is the relative rate of change in the population, compared to the environment. If the environment changes slowly, relative to the process of adaptation, agents will be well-adapted given the constraints. But if the environment changes quickly, relative to the speed of adaptation, then at any given moment most of the relevant population may be poorly adapted.192 Thus an important general argument against institutional analogues of natural selection is that the rate of change in the political and social world is, in general, much faster than in the natural world. Against the background of a natural environment that is relatively stable over periods of millions of years, it is plausible to proceed on the assumption that most extant organisms are well-adapted. In institutional settings such as the market, however, the higher rate of environmental change means that many extant firms may be poorly-adapted at any given time, even if there is strong pressure towards efficiency. An analogous problem arises here. Nothing at all in the demandside models implies, in and of itself, anything about (1) the rate at which the common law converges to efficiency or (2) the rate of change in the background social, economic and policy environment against which efficiency must be judged. If the speed of convergence is slow, relative to the rate of change in the background environment, then even a systematic tendency towards efficiency will not guarantee that most common-law rules are efficient. Indeed, if the rate of environmental change is high, many or most common-law rules might be inefficient at any given time, even if the pressure towards efficiency is powerful. If the common law could adapt to a changing environment at no cost, it would take no time for the common law to reach the optimum, and it would be perfectly efficient at any given time. However, there is a cost of 192
See Jon Elster, Explaining Social Behavior: More Nuts and Bolts for the Social Sciences 280 (2007).
T C C L: E
109
adaptation, usually discussed under the rubric of path-dependence. The term is protean,193 but a simple interpretation in this setting is that there is a positive cost of switching from one legal rule to another, above and beyond the cost of generating rules in the first instance. Path dependence means that where the common law startswhich is typically random, in these modelsconstrains where it can go, and how quickly. As the rate of environmental change becomes more rapid, a given level of switching costs becomes more significant, because it ensures that there will on average be a greater distance between the current location of the common law and its optimal location.
Path-Dependence and the Relative Efficiency of Statutes What of legislation in a changing environment? Does it do any better than common law? The answer is, as always, dependent upon initial assumptions; but some recent models show that under robust conditions legislation will indeed perform better than common law as the pace of environmental change increases.194 Because common law is path-dependent and thus sticky, the more rapidly the environment changes, the greater the comparative efficiency of statute law, even if statute law is less than fully efficient because of the limits of legislators’ foresight or interest-group pressures. The reason is that statutes can innovate more rapidly and completely than the common law, when there is an abrupt change of circumstances. Once the fixed cost of enacting a statute has been paid, there is no extra cost to the legislature of enacting a new rule that is dramatically different than the old rule. Judges, by contrast, face increasing marginal costs for more dramatic departures from precedent, which implies that in a system of precedent incremental change will be the norm.195 Given this, “the fundamental trade-off is between evolution towards efficiency when the social optimum does not change, and rapid legal innovation when it does. . . . [S]tatutory intervention enables the legislature to adapt the law not only to the novel requirements of society, but also to its own biases, or the preferences of the special-interest groups that influence it. This price is 193
194
195
For different senses, see Oona A. Hathaway, Path Dependence in the Law: The Course and Pattern of Legal Change in a Common Law System, 86 Iowa L. Rev. 601 (2001); Paul Pierson, Politics in Time: History, Institutions, and Social Analysis (2004). See Giacomo A.M. Ponzetto and Patricio A. Fernandez, Case Law vs. Statute Law: An Evolutionary Comparison, J. Legal Stud. (forthcoming), available at http://www.people.fas. harvard.edu/~ponzetto/PF.pdf. See id. at 7, 15–16.
110
L L R
only worth paying when social change in sufficiently intense.”196 Casual observation supports this model; “as the pace of technological, economic and social change has accelerated over the last two centuries, so has legislative activity intensified.”197 In Anglo-American jurisdictions, the common law has been gradually supplanted by statutes and area-specific codes.198
Rules, Standards, Legislation, and Common Law How do these points relate to the large literature, in law and economics, on the choice between “rules” and “standards”?199 Rules have more informational content than standards, which are more open-ended. “Bedtime is at 8:00 p. m.” is a rule; “bedtime is when the children are tired” is a standard. Rules are systematically overinclusive and underinclusive with respect to their policy rationales (under the rule, some children will be put to bed when not tired, while some children will not be put to bed although they are tired). Rules require more investment ex ante, because they are costlier to formulate than standards, while standards require more work from decisionmakers ex post or at the point of application, who must directly apply policy rationales to facts with less structured guidance from rules. In one standard model, therefore, rules are more efficient when they govern frequently-repeated and homogenous transactions, because the higher up-front cost of formulating the rule can be spread over more cases, and because the risks of ex ante error are lower. Conversely, standards are more efficient when cases are heterogeneous, because information obtained at the point of application is more important, and when the environment is changing rapidly, because rules obsolesce more quickly.200 For present purposes, the important point is that the choice between rules and standards is orthogonal to the choice between legislation and common law; although related, the two issues are hardly the same. Legislatures can and often do enact standards, while common-law courts can and often do promulgate relatively sharp-edged rules. Moreover, the ex ante/ ex post tradeoff that structures the choice between rules and standards 196 197 198 199
200
Id. at 26. Id. at 27. Id. at 26. See, e.g., Louis Kaplow, Rules Versus Standards: An Economic Analysis, 42 Duke L.J. 557 (1992); Kathleen M. Sullivan, The Supreme Court, 1991 Term—Foreword: The Justices of Rules and Standards, 106 Harv. L. Rev. 22 (1992); Cass R. Sunstein, Problems with Rules, 83 Cal. L. Rev. 953 (1995). See Kaplow, id.; Vincy Fon and Francesco Parisi, On the Optimal Specificity of Legal Rules, 3 J. Institutional Econ. 147 (2007).
T C C L: E
111
can take place on either side of the divide between legislation and common law. The high court of a jurisdiction often faces the choice between formulating a rule or a standard, to govern lower courts in future cases; that high court will face some of the same tradeoffs as a legislature deciding whether to formulate a rule or standard to govern courts. In particular, the relative advantage of standards in a rapidly changing environment is not necessarily, or at all, an institutional advantage of common law over legislation. If rules obsolesce more quickly in such environments, the institutional question is whether legislatures and the executive, on the one hand, or rather courts, on the other, can supply changed rules at lower cost, including the opportunity costs of the time institutions take to adopt rules. If path-dependence and transition costs are high—if common-law precedents are quite sticky—then legislation is likely to be able to supply necessary change more efficiently than common-law adjustment. Furthermore, one must consider not only the rapidity of change, but a closely related issue, the complexity of the environment. If it is true, as I suggested in Chapter 2 and above, that legislatures and agencies have epistemic advantages over courts, then increasing complexity supports a shift of law-developing authority towards legislatures.201
When Was the Common Law Efficient? I have indicated the conditions under which the common law might tend towards or reach efficiency; although it is hard to know in the abstract whether or not those conditions hold, there are no real grounds for confidence that they do. A central point is that the standard demand-side models, even if robust, carry no implication at all that most common-law rules will be efficient, at any given time. One must also consider the rate at which common law tends to efficiency, the rate of environmental change, and the increasing relative efficiency of legislation as the rate of environmental change increases. The implication is that the thesis of common-law efficiency is most plausible in and for periods in which the rate of economic, technological and political change is relatively slow, but is less plausible where change proceeds relatively rapidly especially when the common law is compared to legislation.202 Above, I mentioned the thesis that the common law was efficient into the 19th century but has since ceased to be so, because of the 201 202
See Fon and Parisi, id. at 157–58. See generally Ponzetto and Fernandez, supra note 194.
112
L L R
rise of organized litigation groups or the decline of supply-side competition among legal systems. The current conjecture is in the same family, although the mechanism involves the relative rates of change in the common law and the background environment. Perhaps Burke’s account of the latent wisdom of the common law, interpreted in evolutionary terms, offered a plausible account of common-law efficiency in England in the 17th and 18th centuries, because the common law had time to evolve close to efficiency in a relatively stable environment. Even if so, there is no reason to think that the common law is efficient in other times and places, such as today.
Demand-side Models: Constitutional Implications? Let us turn to the question whether the demand-side models of commonlaw efficiency, even if internally robust, have any payoff for constitutional law. In general, they do not. First, it is familiar that efficiency is a dubious normative goal for constitutional law. Second, the basic demand-side model uses essentially the same mechanism—the willingness of regulated parties to bid more to avoid deadweight losses—that powers the standard model of legislative efficiency. The basic model thus offers no convincing reason to assume that judicial precedents will converge to efficiency while legislation does not. Third, even if legislation has no systematic tendency to converge to efficiency, legislation can respond more quickly to a changing environment. Perhaps the common law is always tending towards efficiency, yet it may nonetheless at any given moment be further away from the optimum than is the body of legislation at the same moment.
Should Constitutional Law Be Efficient? Efficiency is a deeply controversial general goal for constitutional law. I will only touch upon the relevant issues here, both because the point is familiar,203 and because it yields only an external critique of the premises of the evolutionary interpretation of Burke, rather than an internal critique of the interpretation’s logic. As mentioned at the outset of the chapter, one well-known set of problems involves the moral status of efficiency. If efficiency is equated with wealth maximization, there is the major problem that wealth, in itself, is not a value.204 If efficiency is equated with either Pareto optimality 203 204
See, e.g., Elhauge, supra note 188. See Ronald Dworkin, Is Wealth a Value?, 9 J. Legal Stud. 191 (1980).
T C C L: E
113
or satisfaction of the Kaldor-Hicks criterion, under which winners gain more than losers lose, distributive concerns come to the fore.205 These questions about the moral status of efficiency do not apply, at least in the same way, to efficient common law in the subconstitutional sense. Where the subconstitutional common law is efficient, but there are no constitutional constraints, any desired redistribution can take place through the tax and transfer system; doing so will itself be the efficient means of redistribution.206 If constitutional common law is used to block redistributive transfers that are inefficient but (arguably) required by justice, however, then the dubious moral status of efficiency is consequential. One might sidestep or dilute these problems by distinguishing between or among different constitutional provisions. Perhaps efficiency should be the goal of constitutional law under the Takings Clause, but not under the Due Process Clauses. Yet nothing in the evolutionary model of the common law is tied to these textual and doctrinal differences; the relevant mechanisms are not sufficiently fine-grained to accommodate them. Those differences could only be taken into account under an intentional model, in which judges purposefully promote efficiency, either because they have an intrinsic taste for it, or because public pressures force them to do so. An intentional model of this sort, however, would no longer be Burkean or Hayekian in any obvious sense; it belongs to the realm of deliberate social engineering, by judges, that Hayekians and Burkeans deride as “innovation.” Nor, of course, would an intentional model be Darwinian either.
Legislative Efficiency? Suppose, however, that the goal of constitutional law is in general to promote efficient law. Suppose also that a line of constitutional common-law decisions have established a certain rule, and that a course of legislation produces a statute conflicting with the constitutional common-law rule. It does not follow that the judges would best promote efficiency by invalidating the statute on constitutional grounds. The very premises suggesting that the common law tends towards efficiency also suggest that legislation tends towards efficiency.
205 206
For a clear overview of these issues, and of the massive literature, see Matthew D. Adler and Eric A. Posner, New Foundations of Cost-Benefit Analysis (2006). Louis Kaplow and Steven M. Shavell, Fairness Versus Welfare (2002).
114
L L R
Recall that the mechanism driving the basic model of common-law efficiency is that inefficient precedents inflict deadweight losses, and that the parties who suffer those losses will have greater incentives to attempt to overturn them than their beneficiaries will have to defend them (assuming equal stakes). Over time, inefficient precedents will be challenged more often and more frequently overturned. This model is, however, essentially the same as a standard model suggesting that legislation will be efficient, and for the same reasons.207 In this model, “pressure groups” that suffer deadweight losses from inefficient legislation will spend more to overturn such laws than their beneficiaries will spend to defend them. Despite the status quo bias of the legislative process, which favors groups defending inefficient legislation, the systematic tendency over time will be towards an efficient corpus of legislation. The parallel between this model and the basic model of common-law efficiency was clear to all, right from the beginning.208 Both the basic model of common-law efficiency and the standard model of legislative efficiency assume equal distribution of stakes across groups. Accordingly, the standard model of legislative efficiency has often been questioned by public-choice theorists who posit that differential organization by groups with unequally distributed stakes in legislative outcomes; such groups will seek and obtain inefficient legislation, and those who suffer the deadweight losses of such legislation will be insufficiently organized to resist.209 But what is sauce for the goose is sauce for the gander. The same groups will also enjoy their organizational advantages in the adjudicative process,210 including the process of common-law constitutionalism.211 Common-law constitutionalists have attempted to draw nonarbitrary distinctions between legislative and adjudicative processes, in ways that favor the latter. To date none are convincing. One ingenious argument is that the minimum bid—the minimum a group must pay to become a player in the lawmaking game—is lower in the courts than in the legislative 207 208 209 210 211
Gary S. Becker, A Theory of Competition Among Pressure Groups for Political Influence, 98 Q.J. Econ. 371 (1983). See Rubin, supra note 167. See Mancur Olson, The Logic of Collective Action: Public Goods and the Theory of Groups (1965), and a subsequent massive literature. See Elhauge, supra note 188; Frank B. Cross, The Judiciary and Public Choice, 50 Hastings L.J. 355 (1999). See A.C. Pritchard and Todd J. Zywicki, Finding the Constitution: An Economic Analysis of Tradition’s Role in Constitutional Interpretation, 77 N.C. L. Rev. 409, 494–501 (1999) (discussing rent-seeking through Supreme Court litigation and decisionmaking in constitutional cases).
T C C L: E
115
process, while the marginal benefits of incremental expenditures tend to be higher in legislatures than in courts; both points are intended to show that well-funded groups have a greater advantage in legislatures.212 Among other problems with this argument,213 however, the point about marginal benefits of expenditures undermines the basic mechanism that is supposed to drive the common law to efficiency. If litigation expenditures quickly reach a point of zero marginal benefit, then it will not matter that parties suffering deadweight losses would be willing to pay more to challenge inefficient precedents then the beneficiaries would be willing to pay to defend them, because the extra expenditures will have no positive effect on the likelihood of overturning the inefficient precedents.
Efficiency, Social Change, and Thayerism Putting aside the foregoing points, let us suppose now that efficiency is a plausible general goal for constitutional law, that judicial decisions systematically tend towards efficiency, and that legislative decisions do not. It still does not follow that judges will best promote efficiency by invalidating statutes that conflict with a contrary line of constitutional precedents. As the rate of environmental change increases and the stickiness of the constitutional common law increases, the efficiency-maximizing approach is increasingly likely to be Thayerism—judicial deference to legislatures in constitutional cases. If the background environment is changing rapidly, and if the constitutional common law is quite sticky, then legislation in force at any given time will tend to be more efficient than contrary precedent. Although the latter systematically tends towards efficiency at all times, it never attains it at any point, because the environment creates a moving target. New legislation has no dynamic tendency towards increasing efficiency over time (we are supposing), but may simply start at a higher level of efficiency than the constitutional common law rules in place at a given moment, which were developed in a previous environment and now suffer from obsolescence. This is a short-run advantage of legislation that dissipates as the common law evolves towards efficiency in the new environment; but if further environmental change ensues, the common law may 212
213
See Thomas W. Merrill, Does Public Choice Theory Justify More Intrusive Judicial Review After All?, 21 Harv. J.L. & Pub. Pol’y 219, 221–22 (1997); cf. Todd. J. Zywicki, Gordon Tullock’s Critique of the Common Law 46 (George Mason Law & Econ. Research Paper No. 07-13, 2007), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=964781. For further discussion, see Cross, supra note 210.
116
L L R
never catch up. Where environmental change is rapid, a series of short runs is all there is. This is an evolutionary parallel to the Bentham-inspired argument, discussed in Chapter 2, that legislatures systematically tend to have more current information, and thus better information overall, even if judicial decisions more effectively aggregate the information past judges possessed.
Supply-Side Models: The Cardozo Theorem So far, we have been considering evolutionary interpretations of Burke, and in particular models in which the common law evolves toward efficiency—with efficiency understood, as is standard in law and economics, as some version of Kaldor-Hicks efficiency or wealth maximization. As mentioned, however, more recent economic models directly address the putative informational efficiency of the ordinary common law. These models might easily be adapted to support a version of common-law constitutionalism. What follows is therefore in the nature of a preemptive discussion, justified by the close affinity between these models and the Burkean approach. Economists Nicola Gennaioli and Andrei Shleifer have developed illuminating models based upon the “Cardozo Theorem”: “The eccentricities of judges balance one another. One judge looks at problems from the point of view of history, another from that of philosophy, another from that of social utility, one is a formalist, another a latitudinarian, one is timorous of change, another dissatisfied with the present; out of the attrition of diverse minds there is beaten something which has a constancy and uniformity and average value greater than its component elements.”214 As Cardozo’s dictum focuses on the behavior of judges and not that of litigants, Gennaioli and Shleifer use it to develop supply-side models of legal change, in contrast to the demand-side models discussed previously.215 214
215
The Cardozo Theorem is named, and discussed, in Gennaioli and Shleifer, supra note 180. A companion paper is Nicola Gennaioli and Andrei Shleifer, Overruling and the Instability of Law, 35 J. Comp. Econ. 309 (2007). Although Gennaioli and Shleifer do consider how their model of overruling can be extended to incorporate demand-side effects, they do not do so for their model of distinguishing. See Gennaioli and Shleifer, id. at 320–22. In any event, the distinctive and novel feature of their
T C C L: E
117
Their work combines an information-aggregation approach with an evolutionary approach: their premise is that convergence to efficiency occurs in the sense that the legal rules evolve to incorporate all information possessed by the relevant pool of judges. In the best case, the unintended byproduct of countervailing judicial biases is that all material distinctions are built into the body of legal rules, and the law becomes increasingly precise.216 The models are explicitly epistemic. Whatever its merits as a picture of the evolution of ordinary common law, however, the Cardozo Theorem cannot be used to support any robust version of the constitutional common law. (I do not claim that Gennaioli and Shleifer intend or would endorse such a use; indeed, we will see that the implication of their models is that constitutional common law is less likely to be informationally efficient than ordinary common law). First, the internal logic of the models rest upon an excessively sharp distinction between overruling and distinguishing, one that is an artifact of the modeling strategy. Moreover, convergence to informational efficiency, and indeed convergence of any sort, is least likely to occur when the polarization of judicial views is high, perhaps because of the politically controversial character of the subject. And ideological polarization is a chronic circumstance of modern constitutional adjudication, at least at the level of the Supreme Court.
Overruling, Distinguishing, and Constitutional Law To date, the Cardozean models yield different results for two different mechanisms of common-law change—the overruling of cases and the distinguishing of cases.217 Distinguishing, but not overruling, tends towards informational efficiency. In a regime of distinguishing, legal evolution is beneficial on average, because “the informational benefit of distinguishing improves the quality of the law”218—although even under distinguishing a high degree of polarization across judges will dampen the process. In a regime of overruling, the law tends to lurch between polar positions without informational improvement.
216 217 218
approach is the focus on the supply side—on efficiency as the unintended byproduct of uncoordinated decisionmaking by a series of biased judges over time. See Gennaioli and Shleifer, supra note 180, at 62–63. Distinguishing is addressed in Gennaioli and Shleifer, supra note 180. Overruling is addressed in Gennaioli and Shleifer, supra note 214. Gennaioli and Shleifer, supra note 180, at 47.
118
L L R
This divergence results from the analytic setup. In the model of distinguishing, all judges are assumed to be more or less biased in one direction or another—too much liability or too little, relative to optimal liability. When Case 1 is decided in a way that conflicts with the biases of the judge sitting to decide Case 2, the Case 2 judge may distinguish the precedent, but—here is the critical condition—only on “material” or relevant dimensions. If in Case 1 the court held that dog-owners are always liable when strangers are bitten, then in Case 2 the court may distinguish the Case 1 precedent by modifying the rule to say that liability does not attach when the dog was leashed. But the Case 2 court may not say that liability does not attach because the bite occurred on a Tuesday. The result is that distinguishing cases over time adds material dimensions to the law, and this is a kind of extra information that adds precision to legal rules; each judge who distinguishes a prior precedent contributes a portion to the law’s increasing precision over time.219 Under certain conditions, judges’ differing biases wash out and the benefits of increasing precision are all that is left. Most optimistically, the law may even converge to an informationally efficient regime in which rules are maximally precise, so that the social losses from failing to sort between materially unlike cases are minimized. However, we will soon see that the conditions under which this occurs are stringent, and especially unlikely to obtain for constitutional, as opposed to ordinary, common law. The overruling model is different. “Since overruling, unlike distinguishing, does not bring new material dimensions into the law, it leads to the volatility of legal rules without a tendency to improve the law over time. With overruling, there is no benefit of legal evolution.”220 Where polarization among judges is high, the legal rule will never converge to a stable state (let alone an efficient stable state); rather it will fluctuate between extremes. Convergence to efficiency can only occur in the special case in which unbiased judges are more likely to vote to overrule precedents than biased judges and in which the degree of polarization is only intermediate221—an “implausibly stringent” set of conditions.222 The Cardozean models thus draw a sharp contrast between a regime of distinguishing, in which common law evolution is beneficial on average under a wide range of conditions, and a regime of overruling, in which common law is volatile and converges on efficiency only under very 219 220 221 222
See id. at 53–57. See id. at 62–63. See Gennaioli and Shleifer, supra note 214, at 317–18 (Proposition 4). Id. at 323.
T C C L: E
119
unlikely conditions. However, this contrast is an artifact of the modeling setup. In these models, material distinctions count as “distinguishing,” whereas immaterial ones do not; they are automatically classified as a form of overruling. But common-law theory recognizes that the sheer accumulation of material distinctions can itself amount to a form of overruling. Cases are sometimes widely recognized as having been “distinguished to death,” such that all lawyers realize they have been implicitly overruled or “limited to their facts.” Moreover, common-law theorists show that judges often implement their biases precisely by distinguishing precedents based on immaterial or irrelevant dimensions—by holding that dog owners are safe if the bite occurred on a Tuesday.223 In the constitutional common law, Chief Justice William Rehnquist noted the same phenomenon224—and, some say, was himself a master of the quasi-relevant distinction. It is true that from the standpoint of the individual judge, using material distinctions will be strictly preferable to using irrelevant distinctions; the former, unlike the latter, both allows the judge to indulge her biases and also supplies beneficial information. But this holds only “until material dimensions are exhausted.”225 The qualifier is critical. In the constitutional common law, and perhaps also in ordinary common law, the limits of materiality are quickly reached. Consider the bewildering series of distinctions that proliferate in the Supreme Court’s jurisprudence of state sovereign immunity under the Eleventh Amendment,226 where at present states can be sued in federal court for equitable relief but not for damages, unless Congress has clearly said otherwise, but only under its power to enforce the Fourteenth Amendment rather than (say) the Commerce Clause, or unless the United States rather than a private party is the plaintiff, and so on.227 The process that led to this patchwork of rules is one of justices advancing distinctions that were immaterial, given the prior law, and embedding them in the new legal rules. Remarkably, proponents of all competing views of the Amendment seem to agree that the current rules are wrong, given their
223 224 225 226
227
See Karl N. Llewellyn, The Common Law Tradition (1960); Julius Stone, Precedent and the Law (1985). See William H. Rehnquist, The Supreme Court: How It Was, How It Is 302 (1987). Gennaioli and Shleifer, supra note 180, at 61 n.8. I use “Eleventh Amendment” only as a shorthand, bracketing the question whether the Court is here developing a body of law rooted in the amendment’s text or rather in background structural principles of federalism. For an overview of these “complex and often counterintuitive interpretations” of the Eleventh Amendment, see Laurence H. Tribe, American Constitutional Law (3d ed., Volume 1), at 519–66 (2000).
120
L L R
own views; all that keeps the current rules in place, despite their being no one’s first choice, is that there is no agreement on what set of rules should replace them. Perhaps a regime of rules that are everyone’s second choice satisfies as many people as much as possible, and is thus efficient in one sense, but the process of adding cynical distinctions does not plausibly enrich law’s informational content.
Cardozean Constitutionalism? Apart from the internal logic of the models, their main implication is that constitutional common law should be both highly volatile and far from efficient. One key variable is the effect of judicial polarization—the distance between the views of different judges—on the shape and pace of legal change. Cardozo’s hope that “the eccentricities of judges balance one another” and yield a type of “constancy” in the law turns out to be optimistic. As Gennaioli and Shleifer explain, in a regime of overruling, when polarization is high, the common law will simply fluctuate or vacillate from one polar view to another; not only will the law not converge to efficiency, it will never converge at all.228 Vacillation between polar views is a well-known effect of overruling in the constitutional common law. Consider the sequence from National League of Cities v. Usery,229 which announced constitutional protection for the “essential functions” of state governments, to Garcia v. San Antonio Metropolitan Transit Authority,230 which overruled the earlier precedent— with the dissenting justices issuing a warning that the wheel would turn yet again.231 In the event the dissenters, when they obtained a majority, did not reinstate National League of Cities, but pressed forward with similar implicit federalism-protecting doctrines, such as the law of state sovereign immunity discussed above, and the principle that states cannot be commandeered into enacting or enforcing a federal regulatory program.232 In a regime of distinguishing without overruling, Cardozo’s conjecture does better, because legal change through distinctions adds information that overruling does not. Even with high polarization, material distinctions always add precision to the law, and some degree of polarization is affirmatively beneficial, because it induces distinctions in cases where 228 229 230 231 232
See Gennaioli and Shleifer, supra note 214, at 324. 426 U.S. 833 (1976). 469 U.S. 528 (1985). See id. at 580 (Rehnquist, J., dissenting); id. at 589 (O’Connor, J., dissenting). Printz v. United States, 521 U.S. 898 (1997).
T C C L: E
121
unpolarized judges would simply acquiesce in whatever decision was first issued. Although not all extant rules will be efficient at any point, the average tendency of common law over time will be towards efficiency.233 On the other hand, high polarization does reduce the benefits of the distinguishing process, because greater bias causes each judge to use cruder and less informative distinctions.234 When polarization passes a certain threshold, the judge-made law tends less strongly to efficiency and may even become less efficient, because the costs of polarization outstrip the benefits; in other words, there is “an inverted U-shaped relationship between polarization and the efficiency of judge-made law.”235 The upshot is that [t]he evolution of common law would produce most socially efficient results in the areas of law where there is room for change and updating, but where the disagreement among judges is not extreme. The relatively apolitical yet still changing areas of law, such as contract and corporate [law], are the likely candidates for relatively efficient outcomes resulting from the decentralized evolution of judge-made law. In the extremely political areas of law, by contrast, the likelihood that the law gets stuck on a wrong trajectory is higher.236 This conclusion implies that Cardozo’s dictum is least likely to justify the constitutional common law—especially the highly polarizing constitutional common law governing rights and liberties, which is precisely where many common-law constitutionalists want to stake their claims.237 Here some qualifications are necessary. A large set of constitutional controversies are easy, or noncontroversial, or unpolarizing.238 The subset that are litigated, and the sub-subset litigated to the Supreme Court, are more likely to be highly polarizing, although not uniformly so. Moreover, structural constitutional law is often, perhaps on average, less polarizing than the constitutional law of rights and liberties. When all the qualifying is done, however, the comparative point remains: taken together and on average, constitutional cases are more polarizing than the ordinary common law. Relative to ordinary common law, constitutional law is more 233 234 235 236 237 238
See Gennaioli and Shleifer, supra note 180, at 60–62. See id. at 62. See id. Id. at 61–62. See, e.g., Poe v. Ullman, 367 U.S. 497, 543 (1961) (Harlan, J., dissenting). Cf. Frederick Schauer, Easy Cases, 58 S. Cal. L. Rev. 399 (1985).
122
L L R
pervasively concerned with distribution rather than coordination or cooperation; the scope for common interest, somehow defined, in the rules governing contract, property and tort is greater, the scope for common interest in the rules governing abortion, or gay marriage, or free speech in wartime is less. Although I believe this claim will be intuitively plausible to constitutional lawyers, it is hard to find direct evidence either for or against it. However, we may provide indirect support for the intuition by observing that issues are often constitutionalized precisely when, and because, they are polarizing. Partisan entrenchments, in which an outgoing coalition attempts to constitutionalize a favored policy to bind the hands of successors, are routine.239 And nonjudicial institutions will sometimes affirmatively support or encourage the judges to resolve controversial issues through constitutional review,240 precisely because those issues are so polarizing that legislators on all sides hope to duck responsibility or to make a political target of the judges.
Conclusion: Burke, Bentham, and Thayer Revisited A centerpiece of epistemic legalism is common-law constitutionalism. In turn, common-law constitutionalism adopts Burke’s and Hayek’s views about the limits of human reason, and attempts to translate those views to the constitutional setting. By contrast, I have attempted in this chapter and the last to update Bentham’s critique of the ordinary common law, adapting it to the distinctive setting of the constitutional common law. The informational and evolutionary mechanisms canvassed in these chapters lead, not to robust common-law judicial review in constitutional cases, but instead to James Bradley Thayer—to extensive judicial deference to legislative enactments. The very mechanisms that common-law constitutionalists invoke in praise of constitutional precedent cast legislation in a better light still. If many judicial minds are better than one or a few, many legislative minds are plausibly best of all. If the social goal is to maximize the epistemic quality of lawmaking, the best regime will combine deferential judges with common-law constitutionalism by legislatures, who fill constitutional gaps, resolve ambiguities, and adapt constitutional law to new circumstances. 239 240
Howard Gillman, How Political Parties Can Use the Courts to Advance Their Agendas: Federal Courts in the United States, 1875–1891, 96 Am. Pol. Sci. Rev. 511 (2002). See Keith Whittington, “Interpose Your Friendly Hand”: Political Supports for the Exercise of Judicial Review by the United States Supreme Court, 99 Am. Pol. Sci. Rev. 583 (2005).
Chapter 4
Justices and Company “I told you before I’m not a scientist. (Laughter). That’s why I don’t want to have to deal with global warming, to tell you the truth.” Justice Antonin Scalia, at oral argument in Massachusetts v. EPA (Nov. 29, 2006)
“Breyer says that if the only thing that matters is historical truths from the time of the Constitution, ‘we should have nine historians on the court.’ Scalia says . . . that a court of nine historians sounds better than a court of nine ethicists.” Dahlia Lithwick, “Justice Grover vs. Justice Oscar,” Slate (Dec. 6, 2006)
I
now turn from the allocation of lawmaking authority across institutions to the design of the institutions themselves, including the sociological determinants of epistemic performance. An assumption of epistemic legalism is that the courts, especially the Supreme Court, should be staffed entirely by lawyers. I will question this assumption. It focuses myopically on expertise or competence, which is only one component of epistemic performance; another component is epistemic diversity, which can be 123
124
L L R
increased by diversifying the professional composition of the courts. Moreover, even as to competence, the nature of the Court’s docket is such that legal competence is merely one of the bodies of expertise that an epistemically well-functioning Court should possess. On strictly epistemic grounds, rather than the ideological ones he adduced, Bentham was correct that the lawyers’ monopoly of adjudication is socially harmful. In short, I will suggest that we should have lay justices. By “lay justices” I mean justices of the Supreme Court of the United States who are not accredited lawyers. Currently the number of lay justices is zero, although there is no constitutional or statutory rule that requires this. Commentators who urge that the Supreme Court should be diverse on all sorts of margins—methodological diversity, ideological diversity, and racial or ethnic or gender diversity241—say little or nothing about professional diversity on the Court. By contrast, I will claim that the optimal number of lay justices is greater than zero. In the strong form of the argument, it would be a good idea (whether or not it is a politically feasible one) to appoint an historian, economist, doctor, accountant, soldier, or some other nonlawyer professional to the Court. In a weaker form of the argument, it would be desirable at a minimum to appoint more dual-competent justices—lawyers who also have a degree or some other real expertise in another body of knowledge or skill. The topic of lay justices, when it is discussed at all, is usually discussed on populist or jurisprudential premises quite different than the ones advanced here.242 It is no part of my argument that it would be more “democratic” to have courts with some nonlawyers; or that some fraction of the cases that reach appellate courts are pervasively indeterminate;243 or that law, or constitutional law, or Supreme Court constitutional law, is “politics” anyway.244 Similarly, Bentham criticized the legal monopoly of adjudication largely on the ground that the legal profession is a cartel that aims to promote lawyers’ sinister (that is, self-regarding) interests. I disavow that claim as well. 241
242 243 244
For methodological diversity, see Richard A. Posner, Law, Pragmatism, and Democracy 354 (2003); for ideological diversity, see David A. Strauss and Cass R. Sunstein, The Senate, the Constitution, and the Confirmation Process, 100 Yale L.J. 1491 (1992); for racial diversity, see Patrice M. Pitts and Linda Vinson, Breaking Down Barriers to the Federal Bench: Reshaping the Judicial Selection Process, 28 How. L.J. 743 (1985); for gender diversity, see Nicholas O. Alozie, Distribution of Women and Minority Judges: The Effects of Judicial Selection Methods, 71 Soc. Sci. Q. 315 (1990). See the very brief discussions in Fredrick Schauer, Judging in Corner of the Law, 61 S. Cal. L. Rev. 1717, 1731–33 (1988); John Denvir, Proudly Political, 37 U.S.F.L. Rev. 27 (2002). See Schauer, id. See Denvir, supra note 242.
J C
125
My argument, by contrast, assumes that all actors are wellintentioned, in the sense that they are aiming for right answers somehow defined, not for the answers that will benefit them personally or that will promote their ideological aims. I advance an argument based strictly on the epistemic advantages of lay justices. Drawing on important recent work that underscores the epistemic benefits of cognitive diversity in institutional design,245 and on recent work about deliberation, I suggest that a Court with at least some lay justices will reach more right answers across the total set of cases than will a Court with zero lay justices. This is a prime example of the project, laid out in the Introduction, of rehabilitating, recasting, and updating Bentham’s views, on grounds that find support in modern social science and in a resolutely epistemic and technocratic perspective. The reigning assumption has been that, if the analysis proceeds on strictly epistemic grounds, it follows naturally that the legal profession should enjoy a monopoly of the courts generally and of the Supreme Court in particular. Epistemic legalism assumes, and I will assume as well, that lawyers possess a distinctive expertise in “the artificial reason of the lawwhat lawyers know.”246 If so, the thinking runs, lay justices will predictably make many sad blunders and will add nothing positive. But the conclusion does not follow, even if the premise is correct. In strictly epistemic terms, expertise matters, but so does group diversity. In Jury Theorem terms, it is equally critical to have not only highly competent decisionmakers, but also a low or negative degree of correlated biases across the decisionmaking group—and a major source of epistemically desirable diversity is professional diversity.
A Successful Many-Minds Argument? Methodologically, this chapter is crucial to the larger project. If the argument holds, it demonstrates that on an important issue the limits of reason cut in a Benthamite direction—here, against the legal profession’s monopoly of adjudication in the nation’s premier court (and perhaps in lower courts as well, an issue I touch upon at the end of the chapter). But the argument is based on two many-minds mechanisms: the Jury Theorem and the benefits of deliberation among professionally diverse judges. In order to hold, therefore, the argument must surmount the obstacles to
245 246
See Scott Page, The Difference (2007). See Charles Fried, Artificial Reason of the Law or: What Lawyers Know, 60 Tex. L. Rev. 35 (1985).
126
L L R
many-minds arguments identified in Chapter 1. It must pass, that is, through the filters identified there: it must be well-specified, must show that adding lay justices would improve rather than reduce the Court’s net epistemic capacity, must not create self-defeating epistemic bottlenecks or epistemic slack, and must not make false institutional comparisons. I believe the argument does pass through all these filters, but the reader must judge.
Overview To bias the inquiry against lay justices, I assume that law is an objective body of knowledge to which legal training supplies privileged access; that there is a single right legal answer in all cases, even hard cases; and that justices are sincere and vote their best assessment of the legal merits in each case. Even under these assumptions, I suggest, a multimember Supreme Court with some lay justices will do better at reaching right legal answers in some class of cases, and will plausibly do better on net across all cases, than a Court composed solely of lawyer-justices. This is because legal training will cause lawyer-justices to do worse on average at deciding an important class of cases, even according to legal criteria, than nonlawyer justices. At a minimum, a Court with some dual-competent justices will do better in this class of cases than a Court composed solely of pure lawyers. The relevant class is made up of cases in which law itself requires that judges make decisions based in part on nonlegal knowledge. There are two subclasses of cases within this class: (1) cases in which law draws upon specialized knowledge that is not itself legal, such as economic or medical or military expertise; (2) cases in which law draws upon knowledge that is neither specialized nor legal, such as knowledge of “the mystery of human life”247 or of “evolving standards of decency.”248 In the former subclass, nonlawyers bring to the bench distinctive expertise that can make decisions better. In the latter subclass, lawyers have systematic and correlated biases induced by common professional training, but a bench composed of both nonlawyers and lawyers will have uncorrelated or negatively correlated biases, or at least a lower degree of positive correlation. In Jury Theorem terms, the aggregate decisionmaking competence of the group will improve if the Court contains at least some lay justices, while in deliberative terms, nonlawyers will dampen harmful groupthink.
247 248
Planned Parenthood v. Casey, 505 U.S. 833, 851 (1992). Roper v. Simmons, 543 U.S. 551, 561 (2005) (quoting Trop v. Dulles, 356 U.S. 86, 100–101 (1958)).
J C
127
There are tradeoffs to be made, because lay justices will do worse than lawyer-justices in cases where specialized legal knowledge is all that matters. But no sensible guess about the shape of the tradeoffs would suggest that the optimal number of lay justices is zero. A well-motivated appointer—a kind of benevolent planner for the judiciary—would appoint more than zero lay justices, in part to diversify and thereby hedge our bets in the face of uncertainty and disagreement about how a wellfunctioning Court should be composed. The costs and benefits sketched here apply, in dilute form on both the cost and benefit side, to dualcompetent justices. I will begin with some factual background and will then adopt an array of restrictive assumptions, deliberately biased in favor of a bench exclusively composed of pure lawyer-justices. The core of the discussion identifies the major tradeoff: although a bench composed solely of pure lawyers will minimize judicial error in cases in which law draws solely upon specialized legal knowledge, a bench with some lay members will do better at minimizing errors when law draws upon knowledge that is either non-specialized or specialized but non-legal. So long as the latter two categories are substantial and as I will show, they are in fact very substantial—a multimember court that includes more than zero lay members is likely to minimize errors across the total array of cases. At a minimum, dual-competent justices would improve on the current Court, which is dominated by pure lawyers. Subsequently, I examine an institutional alternative: the bodies of knowledge that drive the argument can be incorporated into judicial decisionmaking through deference to administrative agencies, special masters, and juries, and through the briefing and argument of parties and amici curiae. This is a question about whether people with the relevant bodies of knowledge should be brought inside the boundaries of the judicial firm; it is somewhat analogous to the make-or-buy decision that private-sector firms face. Under conditions of high transaction costs, I will suggest, it is better to appoint people with the relevant knowledge to the bench than to rely on case-by-case incorporation of knowledge. On this view, expertise in domains outside of law should be institutionalized through a kind of representation, by putting at least some nonlawyer justices or dual-competent justices on the Court. I will also examine endogeneity issues, such as the effect of lay justices on the Court’s agenda and on the content of the legal rules it creates, and the question whether lay justices would be excessively deferential to lawyer-justices or vice versa.
128
L L R
Some Facts and Assumptions Facts Lay judges are hardly novel. Many legal systems today use lay judges at some levels of the judicial hierarchy. I will briefly survey some major liberal democracies and other nations, but we should be aware of subtle institutional factors that make comparison difficult. Chief among these are that in many legal systems law is not taught in separate professional schools, judging is a separate career from advocacy rather than a later stage in the advocate’s career, and judges may by the terms of their tenure, position, and promotion be more analogous to Anglo-American administrative officials than to federal Article III judges or justices. Still, some sense of the comparative landscape helps to put matters in perspective. In the United States, lay judges of various sorts sit in some 40 states, although usually on low-stakes matters, and in some cases subject to de novo review by a lawyer-judge.249 In the United Kingdom, lords without legal training heard appeals in the House of Lords until 1834;250 today, lay magistrates or justices of the peace hear certain classes of civil and criminal cases.251 In Germany, lay judges sit on all courts of first instance that have criminal jurisdiction; in the Amtsgericht, which hears less serious cases, one professional judge sits with two lay judges, whereas in the Landgericht, which hears all other criminal cases, three professional judges sit with two lay judges. Japan recently introduced lay judges into its criminal justice system. Criminal cases are heard by a panel composed of three career judges and six lay judges, but the career judges cannot be narrowly outvoted; when the decision is made by a vote of five to four, the majority must include at least one career judge. It is sometimes assumed that lay judges are only used on (1) courts of first instance and (2) in low-stakes matters, such as might be handled by a justice of the peace. However, a surprising number of nations use nonlawyer judges on appellate courts. These include Austria, the Czech Republic, France, Italy, and Sweden, as well as Cuba, Kosovo, and Latvia. The last three nations go further and put nonlawyer justices on the highest appellate court, while in France, nonlawyers serve on the Conseil Constitutionnel, 249
250 251
Lay judges are not present in the United States federal system, but are present in state limited jurisdiction courts in about 40 states. See Bureau of Labor Statistics, Judges, Magistrates, and Other Judicial Workers, http://www.bls.gov/oco/ocos272.htm (last visited Feb. 28, 2008). House of Lords Library, Library Note: The Appellate Jurisdiction of the House of Lords, http:// www.parliament.uk/documents/upload/HLLAppellate.pdf (last visited Feb. 28, 2008). See OSCE Office for Democratic Institutions and Human Rights, Legislation Online: United Kingdom, http://www.legislationline.org/?tid=1&jid=54&less=false (last visited Feb. 28, 2008).
J C
129
the highest constitutional authority.252 Most nations, however, have a legal requirement or an implicit practice that allows only trained lawyers to serve on appellate courts of last resort. In the United States, no nonlawyer has ever served on the federal Supreme Court in the modern era, although no rule of constitutional or statutory law excludes them. The qualifier “in the modern era” refers to the era of accredited law schools and legal requirements that students attend such a school, requirements that became universal around 1950. Even before the modern rules took hold, however, every person nominated to the Supreme Court had received some legal training, perhaps by way of apprenticeship or study towards a bar exam. In that sense, we can drop the qualifier and just say that no nonlawyer has ever served on the Court.253 The Court’s history conspicuously lacks any justices who were soldiers, engineers, professional politicians, historians, philosophers, economists, accountants, or doctors and who were not also lawyers. (Later, I will examine the possibility that one can have one’s cake and eat it too, simply by appointing to the Court lawyers who also have expertise or experience in other professions or bodies of knowledge). All this suggests a type of constitutional argument from tradition. Perhaps the global tendency to exclude nonlawyers from the highest appellate courts—although France is an important exception—and the longstanding American pattern that no nonlawyer has ever sat on the Supreme Court, together supply a good normative reason to exclude nonlawyers from the Court in the future. Traditionalists, Burkeans, and Hayekians of various stripes will tend to assume that there is some latent wisdom or epistemically valuable aggregated judgment in the repeated decision of many presidents over time never to appoint a nonlawyer, and in the parallel practice of most other countries to exclude nonlawyers from their high courts. On the other hand, as I have emphasized in previous chapters, there are many longstanding behavioral regularities that have no latent wisdom behind them at all and that lack any epistemic value whatsoever; this may well be one of them. Perhaps nonlawyers have never been appointed to the Supreme Court, in the United States anyway, just because most people have thought, stupidly, that what had never been done in the past must be illegal or harmful. Paradoxically, as I have also emphasized, 252 253
See John Bell, Principles and Methods of Judicial Selection in France, 61 S. Cal. L. Rev. 1757, 1763 (1988). By contrast, “in the early national period nonlawyers served on state supreme courts.” John V. Orth, Who Judges the Judges?, 32 Fl. St. U. L. Rev. 1245, 1248 n.16 (2005).
130
L L R
those who rely upon tradition or longstanding practices as the basis for forming their own judgments about matters of this sort do not contribute anything to the epistemic value of the practice because they are not forming their own judgments independently. Behavioral regularities may rest on various forms of herding or cascades that deprive them of epistemic import. Moreover, if the longstanding practice of appointing only lawyers to the Supreme Court is to be taken seriously, one must hold the view that of the 110 people who have served on the Court, not one should have been a nonlawyer, and that at present, not one of the nine justices should be a nonlawyer. This is an extreme solution that might happen to be right, but extreme solutions are presumptively suspect in complex institutions whose personnel must be chosen under conditions of uncertainty. Indeed, the very unanimity of the practice makes it suspect. Rather than suggesting latent wisdom, it may suggest that the practice rests on a kind of thoughtlessness, such that no one has considered and rejected the arguments for professional diversity on the Court.
Assumptions Discussions of the optimal composition of the Supreme Court tend to bog down in questions about the nature of law, the nature of the appellate caseload, the normative theory of adjudication, and the positive question of what judges maximize. I sidestep these issues by adopting artificial assumptions that are maximally biased in favor of pure lawyer-justices. The assumptions are that law constitutes an objective body of knowledge; that professional training confers distinctive expertise in that knowledge; that all cases have right answers; and that judges are (1) sincere and (2) vote their view of the legal merits. If the optimal number of lay justices is greater than zero even under these restrictive assumptions, it will be even larger once some or all of the assumptions are relaxed. I also assume that all current rules regarding the Court—the tenure of its members, the process by which they are selected, and the voting rules they use—are held constant. A few comments on each assumption follow.
The Objectivity of Law I assume that law consists of a body of observer-independent knowledge in the same way, or to the same degree, as medicine or military science or accounting. There are many other conceptions of objectivity; I have chosen a minimalist one that is plausible, mundane, and sufficient to get the issue off the ground.
J C
131
Professional Expertise I assume that legal training confers expertise. Lawyers—using “lawyer” in its modern sense of “a person who has attended an accredited law school and been admitted to the bar”—know more about the objective body of knowledge called law than do nonlawyers; they are more likely to get the legal answer right.
Right Answers I assume that there are right answers in every case, even the hard cases that come before the Supreme Court. Even if different sources of law point in different directions, there is a single internal legal answer to the question what law requires, all things considered.254 This right-answer assumption is compatible with, but does not require, the further view that the right answer is the one that combines “fit” with “justification” in some suitably weighted function.255 What it does rule out is the idea that law is indeterminate, because legal sources run out or hopelessly conflict with each other, in some fraction of the cases that reach courts of last resort, especially the Supreme Court. This assumption biases the inquiry against lay justices because it blocks an important argument: that, where legal sources run out or conflict, the distinctive legal expertise of lawyer-justices drops out of the picture.256 I hope to show that the argument from indeterminacy picks an unnecessary fight with the right-answer view. Even from within that view, there is an epistemic argument for lay justices because often enough the legally right answer itself incorporates knowledge from other disciplines or from general social life.
Judicial Motivation There is a complex of issues surrounding judicial motivation. Economists ask what judges maximize, although this assumes part of the conclusion by assuming that the motivational structure of judges is consequentialist. The most prominent economic description of the judicial utility function is mushy: it is said to include promoting good legal policy, obtaining the regard of professional and social peers, enjoying leisure, and enjoying the intrinsic utility that arises from playing the law game
254 255 256
See Ronald Dworkin, Hard Cases, 88 Harv. L. Rev. 1057 (1976). See Ronald Dworkin, Law’s Empire (1986). Cf. Schauer, supra note 242.
132
L L R
according to its rules—with the relative weights of these utility-components unclear.257 In legal theory and political science, there are various explicit and implicit models of the motivation of judges, usually appellate judges. On one view, judges vote their conceptions of what law requires; in another, they vote for the policies they think best;258 in yet another, their primary motive is to appeal to relevant audiences, of lawyer, law professors, social groups, and elite opinion-makers.259 Any of these views may further be cast in sincere or strategic terms; the distinction cuts across the different firstorder motivations.260 A judge might, for example, act strategically in order to maximize the chances that his (sincere) view of what the law requires will actually be adopted by the court and will not be overridden by other institutions. More commonly, political scientists model judges as both strategic and interested solely in advancing their views of good policy. On the level of first-order motivations, I will assume that all justices are solely motivated by what, in their best judgment, the law requires— although in some cases the law itself requires that justices defer to the judgments of other decisionmakers or draw upon bodies of nonlegal knowledge, specialized or unspecialized. This assumption biases the inquiry against lay justices, in the sense that it blocks the easy argument that because justices are just doing policy anyway, whether or not legal sources are indeterminate, their legal training is irrelevant. I will also assume that all justices vote sincerely, not strategically. Given the assumption about strictly legal first-order motivations, it is unclear whether a sincere or strategic model is biased against lay justices; I adopt the assumption of sincerity because it simplifies the discussion, without obvious consequences for the argument.
Other Rules Held Constant Many discussions of the Court’s optimal composition slip into confusion because assumptions about other dimensions of institutional design that indirectly affect composition—such as the justices’ tenure, or the voting 257 258 259
260
See Richard A. Posner, What do Judges and Justices Maximize? (The Same Thing Everyone Else Does), 3 Sup. Ct. Econ. Rev. 1 (1994). See Jeffrey A. Segal and Harold J. Spaeth, The Supreme Court and the Attitudinal Model Revisited (2002). See Lawrence Baum, Judges and Their Audiences: A Perspective on Judicial Behavior (2006) [hereinafter Judges and Their Audiences]; Lawrence Baum, The Puzzle of Judicial Behavior (1997); Frederick Schauer, Incentives, Reputation, and the Inglorious Determinants of Judicial Behavior, 68 U. Cin. L. Rev. 615 (2000). See Baum, Judges and Their Audiences, id.
J C
133
rules they use, or the scope of their jurisdiction—are left unclear, or change implicitly as the discussion unfolds. I will hold constant all other institutional features of the Supreme Court as it currently is; in particular, I assume simple majority voting on all merits questions. At the end of the chapter, however, I will relax the assumption that the Court’s docket is constant and will ask whether the introduction of at least one lay justice would have significant (and perhaps undesirable) effects on the type or amount of cases that the justices hear.
Political Constraints Ignored I assume away political constraints. That is, I ask what a well-motivated appointer seeking an optimal composition for the Court would do, regardless of political acceptability. However, I emphatically do not assume away other constraints, such as the cost of information. As discussed below, uncertainty about the Court’s optimal composition is one of the main considerations that militate in favor of appointing at least one lay justice. This combination of assumptions—well-motivated officials, but informational uncertainty—is routine in welfare economics and other disciplines that address optimal policy and institutional design.
Lay Justices as Error-Minimizers With these assumptions in place, the main argument for lay justices focuses on the cost of errors. The cost of reaching decisions, holding constant the cost of error, is discussed subsequently. Error occurs when the justices, acting by simple majority voting on the merits of cases, decree a legal answer that is not the right legal answer, which by assumption always exists. The argument does not require that we be able to say what it is that makes the right answer right, or what the right answer actually is in any given case. All we need say is that the right answer, whatever it is, has certain properties such that certain types of decisionmakers—namely nonlawyers—are more likely to reach it, on average, than are other types of decisionmakers—namely lawyers—across a given set of cases. The argument for lay justices then takes two basic steps. The first step is that, within the total pool of cases the Supreme Court considers, there is a certain class of cases with the following property: nonlawyers are more competent decisionmakers, with respect to that class of cases and on average, than are lawyers. If this is so, then adding at least one lay justice would reduce the costs of error in that class of cases, which is a benefit. The second step is that the benefit would be greater than any costs that would
134
L L R
occur because the participation of at least one nonlawyer means that the justices as a group commit more errors in another class of cases—those in which lawyers are, on average, better decisionmakers than nonlawyers. Two types of many-minds mechanisms, introduced in Chapter 1, underpin the argument: the aggregation of information held by justices through collective voting, and the aggregation of perspectives through deliberation within the Court. The effects of deliberation on group decisionmaking competence are complex and ambiguous, but in an important range of cases, adding at least one lay justice can be expected to improve the Court’s deliberations, when deliberation occurs at all. Experts in outside bodies of knowledge can make self-confirming contributions, can block a group of lawyer-justices from herding towards the wrong answer, and more generally can dilute the likemindedness that arises from common professional training and that is a standard cause of deliberative failure. Subsequently, I also offer the weaker argument that at least some of the advantages of lay justices can be captured by justices with dual competence—justices with knowledge of both law and some other professional discipline. The argument is weaker because it can be accepted even by those who believe that legal training is an indispensable qualification for all justices. It is still highly consequential, however. At present, legal training is not only necessary for all justices but sufficient for all justices, whereas under a regime of dual competence at least some of the Court’s members would be required to possess expertise in other disciplines as well.
Definitions We must begin by defining some useful terms. First, we must roughly distinguish two types of cases—those in which law is autarkic and those in which it is not. In autarkic cases, law is self-sufficient and strictly inwardlooking. The legal right answer can be found solely through the resources of legal texts, history, precedents, and interpretive maxims. The case may be “hard” in either the ordinary-language sense or in the jurisprudential sense—for example, the relevant legal sources may point in different directions, or contain internal conflicts or problems of some other type—but the law does not ask the judge to do anything but to combine these sources into a decision through some decision rule. In nonautarkic cases, law is outward-looking. The legal right answer itself incorporates by reference nonlegal domains of knowledge, which can be more or less specialized—on a continuum running from professional accounting standards, on the one hand, to “evolving standards of
J C
135
decency,” on the other. This incorporation by reference does not make the right answer any less legal or make the law indeterminate; nor does it mean that law has “run out.” It just means that for various reasons, law decides that law should defer to some other domain of knowledge. Whatever is the right answer in that domain of knowledge just is the right answer in law as well. This distinction is just a heuristic for ease of exposition. Perhaps many or most cases at the Supreme Court level are mixed, containing some thoroughly inward-looking legal questions or aspects and some outward-looking questions or aspects. Nothing of substance turns on whether we talk about types of cases or aspects of cases, except that the latter is more complicated, so I will adopt the former mode. Cases in which law is outward-looking can in turn be subdivided into two categories. These two categories are just areas on a continuum, but the distinction will shortly become useful in a rough way. The first category involves cases in which the nonlegal knowledge that law incorporates by reference is specialized, meaning that professionally trained decisionmakers (economists, accountants, historians, military scientists, or doctors) are more likely to reach right answers about questions in that domain. For some recent examples from the Supreme Court, consider the following cases, organized by the type of specialized nonlegal knowledge the justices attempted to bring to bear:
Economics and Finance In Till v. SCS Credit Corp.,261 the justices had to choose one of four possible interest rates to govern “cramdown” proceedings in bankruptcy. Rejecting the “coerced loan” approach, the “contract rate” approach and the “cost of funds” approach, the Court opted for the so-called “formula” approach, on economic and institutional grounds.
Environmental Science and Ecology In South Florida Water Management District v. Miccosukee Tribe of Indians,262 the Court refereed a dispute between a federal water-management project and an Indian tribe. Although remanding for further lower-court findings on some issues, it did not hesitate to offer judgments of fact and policy about some of the disputed waterways: “The District Court correctly characterized the flow through S.9 as nonnatural, and it appears that if 261 262
541 U.S. 465 (2004). 541 U.S. 95 (2004).
136
L L R
S.9 were shut down, the water in the C.11 canal might for a brief time flow east, rather than west. But the record also suggests that if S.9 were shut down, the area drained by C.11 would flood, which might mean C.11 would no longer be a distinct body of navigable water, but instead part of a larger water body extending over WCA.3 and the C.11 basin. It also might call into question the Eleventh Circuit’s conclusion that S.9 is the cause in fact of phosphorous addition to WCA.3.”263
Health and Medical Science In Merck KGaA v. Integra Lifesciences I, Ltd.,264 the Court decided that “[t]he use of patented compounds in preclinical studies is protected under [federal drug statutes] at least as long as there is a reasonable basis to believe that the compound tested could be the subject of an FDA submission and the experiments will produce the types of information relevant to an IND or NDA [that is, applications to market new drugs].”265 Along the way, the Court opined on the nature of “basic scientific research,” its relationship to FDA testing, and the “reality” of the scientific process in the contemporary pharmaceutical industry.
Military Science In United States v. Virginia,266 the Court said that the Virginia Military Institute (VMI) could not exclude women from its officer training program. VMI relied, in part, on a claim that its method of “adversative training” would be undermined by admitting women. The Court asserted that “neither the goal of producing citizen-soldiers nor VMI’s implementing methodology is inherently unsuitable to women.”267
History Any number of cases could supply the example, especially cases in which the Court draws upon the “original understanding” of constitutional texts or upon longstanding traditions. Sometimes, however, the legal analysis becomes even more narrowly focused on the history of particular events. Consider Justice John Paul Stevens’ analysis, in Hamdan v. Rumsfeld,268 of the events behind Ex Parte Quirin,269 a World War II–era precedent that 263 264 265 266 267 268 269
Id. at 97. 545 U.S. 193 (2005). Id. at 208. 518 U.S. 515 (1996). Id. at. 520. 126 S. Ct. 2749 (2006). 371 U.S. 1 (1942).
J C
137
Stevens attempted to debunk by suggesting that the Roosevelt Administration had pressured the Court into reaching a favorable decision.270
Specialized and Nonspecialized Knowledge In these and many other cases, Supreme Court law is outward-looking and incorporates by reference specialized nonlegal knowledge. In an ultimate sense the legal question is one of statutory or constitutional interpretation. But the correct interpretation—what the constitutional or statutory provision is best taken to say—itself depends on what the body of specialized outside knowledge indicates. A second and different category involves cases in which the nonlegal knowledge that law incorporates by reference is nonspecialized, in that nonprofessionals are no less likely to get the right answer; examples are cases in which questions of general social values are concerned. Here too, I assume some form of metaethical realism and cognitivism, such that there can be right answers to questions of general values; what defines the category is that professional legal training confers no special advantage in finding those answers, even though they are, by incorporation into law, the right legal answers. Recent examples from the Court include Roper v. Simmons,271 in which the Court held that “evolving standards of decency” prohibit the execution of juveniles; Atkins v. Virginia,272 in which the same evolving standards were held to prohibit the execution of the mentally retarded; and Lawrence v. Texas, where the Court invalidated laws against homosexual sodomy on grounds that “involve[d] liberty . . . in its more transcendent dimensions.”273
Lay Justices, Expertise, and Bias With this taxonomy in hand, we can turn to substance. The first step in the epistemic argument for lay justices is that under very plausible conditions, a Court that contains at least one lay justice will commit fewer errors in the class of cases in which law is not autarkic. Begin with the easiest case: where the requisite nonlegal knowledge is specialized and the lay justice is a specialist in the relevant domain. For concreteness, let us suppose that the case is about taxation and involves complex accounting and that the 270 271 272 273
See Hamdan, 126 S. Ct. at 2776. 543 U.S. 551 (2005). 536 U.S. 304 (2002). 539 U.S. 558, 562 (2003).
138
L L R
lay justice is a professional accountant. Under these conditions, we must compare the performance of a Court composed of eight lawyer-justices and the professional accountant, on the one hand, with the performance of a Court composed of nine lawyers, on the other. The Jury Theorem, we have seen, states that where right answers exist and where the average competence (likelihood of producing a correct answer) of the decisionmakers exceeds .5, the likelihood that a majority vote of the group will produce the right answer approaches 1 as the group becomes larger. The Theorem does not apply only to factual or causal questions; it can apply whenever there is an exogenously defined right answer, and we are assuming that this is true with respect to all cases, even hard cases. As discussed in Chapter 1, the Jury Theorem need not assume such an exogenous or preference-independent right answer, but can do so. Again, I will assume the existence of such an answer in all cases in order to bias the inquiry against lay justices. Holding constant the number of the Court’s members, adding the accountant-justice should then straightforwardly improve the Court’s performance in the relevant subclass of cases—here, tax cases. Of course there is internal disagreement within other disciplines, just as there is within law. Perhaps the lay justice will take a position on relevant questions that would be controversial within his or her own professional group. But this is somewhat beside the point, because it fails to make the proper comparison. The group is at least as well off with an expert who takes a controversial position as it is with no expert at all. On questions where there is a consensus within the relevant nonlegal profession—questions at the disciplinary core, rather than on the margins or cutting-edges where disagreements arise—the expert will help the group to do much better than would a group with no relevant expertise. On average, adding an expert with relevant specialized knowledge raises the average competence of the group and improves its performance. This case is seemingly easy because we have built in the assumption that there is a match between the requisite nonlegal specialized knowledge here, accounting—and the special skill of the nonlawyer justice—here, an accountant. However, this assumption is dispensable. There is no reason to think that the accountant-justice will do worse than the lawyer-justices, on average, in cases that do not involve accounting but that do involve some other domain of nonlegal specialized knowledge. Adding the accountant-justice is then a pure improvement because it helps in some identifiable subclass of nonlegal specialized cases and does not hurt in other subclasses of such cases.
J C
139
To clarify this point, consider a portfolio of ten cases to reach the Court. In seven of these cases law is strictly autarkic; in one it incorporates by reference accounting knowledge, in one medical knowledge, and in one military science. The Court with one accountant will, as above, be more likely to get the accounting case right. It will be no more likely than the Court with nine lawyers to get the medical and military cases wrong. (Below, I will go further, suggesting that the group with at least one lay justice will in fact be more likely to get all ten cases right; but for now I need not advance that claim). So the presence of the accountant is a pure improvement, so far as the nonautarkic cases are concerned. Of course, it would also have been a pure improvement to add one doctor or one military scientist instead of one accountant. That point is not inconsistent with the expertise argument. It suggests, if anything, that the Court should contain more than one lay justice; a diversity of nonlegal expertise would ensure that there is one specialist in every major subclass of case in which law is not autarkic. I return to this point shortly.
Correlated Bias and Nonspecialized Knowledge What about the subclass of nonautarkic cases in which the requisite knowledge is not specialized—where law draws upon community standards, or asks about “evolving standards of decency,” or incorporates some account of “liberty in its more transcendent dimensions”? In such cases, it is tempting to conclude that adding lay justices cannot make things better. After all, because the relevant issue is not specialized, all people who participate in the mores or culture of the political community have equal access to the facts or values on which the requisite judgments are based. And lawyers are people, too. It is sometimes suggested that if law is “just politics” or “just value judgments,” then justices might as well be elected,274 or should at least be nonlawyers.275 I have put aside this class of arguments, but it is worth pointing out that the suggestion falls flat in any event. Because lawyers are people, too, there is no obvious advantage, from the standpoint of assessing community values, in replacing lawyers with nonlawyers. So runs the tempting line of argument. However, I believe this argument fails, on two grounds. For one thing, even if it were correct, it would
274 275
See Dennis B. Wilson, Electing Judges and Justices: Should the Supra-Legislators be Accountable to the Voters?, 39 Creighton L. Rev. 695 (2006). See Denvir, supra note 242.
140
L L R
not show that there is no improvement from adding lay justices. It would show only that the improvement comes only in the set of specialized nonautarkic cases, while judicial performance in the set of nonspecialized nonautarkic cases would be made neither better nor worse. After all, if lawyers are people, then doctors and economists and military officers surely are as well, and will do no worse on average than lawyers in assessing the transcendent dimensions of liberty. More importantly, however, the argument fails on its own terms. There is a crucial sense in which lawyers are not just like ordinary people. They are part of an elite professional class, and their legal training gives them distinctive professional biases,276 or is at least correlated with such biases,
276
This claim is broadly supported by a large literature in psychology, sociology, and legal ethics on the effect of legal training on lawyers’ values and decisionmaking. For overviews, see, e.g., Susan Swaim Daicoff, Lawyer, Know Thyself (2002); James M. Hedegard, The Impact of Legal Education: An In-Depth Examination of Career Relevant Interests, Attitudes, and Personality Traits Among First-Year Law Students, 4 Am. B. Found. Res. J. 791 (1979); Jon P. Plumlee, Lawyers as Bureaucrats: The Impact of Legal Training in the Higher Civil Service, 41 Pub. Admin Rev. 220, 221 (1981). It is clear from this literature that lawyers are, compared to people generally, more introverted, more rational as opposed to emotional, more judgmental, more competitive, aggressive, and materialistic. See Daicoff, id. at 40–42. It is harder to extract from this literature clear predictions about lawyers’ policy biases. Three relevant claims, in roughly descending order of rigor, are that: (1) legal training gives lawyers a strong status quo orientation and a bias to conventional morality, as compared to similarly educated adults— see Lawrence J. Landwehr, Lawyers as Social Progressives or Reactionaries: The Law and Order Cognitive Orientation of Lawyers, 7 Law & Psychol. Rev. 39 (1982), and June Louin Tapp and Felicia J. Levine, Legal Socialization: Strategies for an Ethical Legality, 27 Stan. L. Rev. 1 (1974); (2) legal training reduces law students’ general concern for social justice—see, e.g., Susan Kay, Socializing the Future Elite: The Nonimpact of Law School, 59 Soc. Sci. Q. 347 (1978)—and reduces their interest in public service or public-interest lawyering—see Robert Granfield, Learning Collective Eminence: Harvard Law School and the Social Production of Elite Lawyers, 33 Soc. Q. 503 (1992); and (3) legal training causes lawyers to favor cumbersome and complicated processes for generating policy—see, e.g., Jerold S. Auerbach, A Plague of Lawyers, Harper’s, Oct. 1976, at 37. The relevant literature is not unanimous on these issues. In tension with claim (1) above, see Thomas Willging and Thomas Dunn, The Moral Development of the Law Student: Theory and Data on Legal Education, 31 J. Legal Educ. 306, 329 (1981) (finding that law student’s moral views are similar to those of the general population and those of graduate students in other professional schools); Susan Daicoff, (Oxymoron?) Ethical Decisionmaking by Attorneys: An Empirical Study, 48 Fla. L. Rev. 197 (1996) (finding that attorneys employ casuistic reasoning to resolve professional ethical dilemmas, rather than relying exclusively on preestablished rules). For a qualified critique of finding (2) above, see Howard S. Erlanger and Douglas A. Klegon, Socialization Effects of Professional School: The Law School Experience and Student Orientation to Public Interest Concerns, 13 L. & Soc’y Rev. 11 (1978) (finding that law students’ attitudes do change away from “social reform” to a “business orientation,” but that the change is small). I am not aware of any rigorous treatment of claim (3) above, which is intuitively plausible but no more. Despite this residual uncertainty, the weight of the literature supports the general and rather unsurprising claim that lawyers as a group display predictable traits and policy preferences that distinguish them both from society at large and from other professions.
J C
141
which may also arise from self-selection into the legal career by certain types.277 The argument for lay justices is the same in either case; what matters is that lawyers as lawyers have a common outlook distinct from other professionals. If lawyers’ values predictably have “the smell of the lamp” about them,278 there will be a systematic bias in the way that lawyers perceive “evolving standards of decency,” and so on. Again the point is emphatically not that law is just politics or anything like that. The point is that because of their systematic professional biases, a group of lawyers will be more likely to make a genuine mistake about what the law itself requires, in the subclass of cases in which law draws on nonspecialized and nonlegal knowledge. This is a version of Bentham’s complaint about lawyers, but translated into epistemic terms. In the language of the Jury Theorem, as discussed in Chapter 2, the crucial point is that lawyers have correlated or systematic biases. Few voters are perfect; what matters is that when voters err, they are as likely to err in one direction as another. In the best case, the uncorrelated biases of different voters will randomly offset each other and wash out at the aggregate level. Recall that, in general, where voters have positively correlated biases, the group judgment will be less accurate, holding average competence constant;279 moreover, the chance of a majority selecting the right answer varies inversely with the correlation of bias across the group.280 This implies that a group of given average competence may do worse than a group with lower average competence but less correlation of biases. Indeed, adding worse-than-random guessers to a group can improve overall performance if their biases are negatively correlated with those of highly competent experts.281 Where this effect occurs, it is because “the uninformed voters drive the average correlation down, thus more than compensating for their relative ignorance.”282 By reducing the correlation of biases, adding lay justices into a group of lawyers can improve the decisional 277 278 279 280 281 282
For a careful attempt to disentangle training effects from the effects of self-selection into the legal career, see generally Daicoff, id. at 68–77. John Hart Ely, Democracy and Distrust: A Theory of Judicial Review 59 (1980). Krishna Ladha, The Condorcet Jury Theorem, Free Speech, and Correlated Votes, 36 Am. J. of Pol. Sci. 617, 625–27 (1992). Id. at 632. Id. at 628. Id. at 629. In Ladha’s example, a group of three decisionmakers each competent at the .8 level cannot be certain of getting the right answer whatever the voting rule. However, if the group adds two uninformed, worse-than-random guessers, whose competence is a mere .3, then the group is certain to choose the right answer so long as the guesses of the uninformed are negatively correlated with those of the informed. See id.
142
L L R
accuracy of the group even as to questions on which none of the group’s members has distinctive expertise. As compared to a Court of nine lawyers, the Court of eight lawyers and one accountant might do better even on questions outside the professional expertise of anyone in the group— questions say, about what “evolving standards of decency” require— because the latter decisionmaking group displays a lower correlation of biases than does the former group. Indeed, these points hold not only for the cases in which law is nonautarkic and draws upon nonspecialized knowledge of community values; it also holds for cases in which law is nonautarkic and draws upon specialized knowledge outside the relevant expertise of the lay justice. Recall the portfolio of ten cases in which seven are autarkic, one involves accounting, one medicine, and one military science. Suppose there is one lay justice, an accountant. Because the presence of this lay justice reduces the correlation of biases, relative to a court of nine lawyers, the Court as a whole may guess better even in the medical and military cases—if those cases present issues as to which lawyers have common biases arising from their professional training. Finally, consider the hypothesis—unlikely but often advanced by lawyers—that lawyers are better at deciding questions outside their expertise than are members of other professions at deciding questions outside their expertise. The counterintuitive finding discussed above shows that this hypothesis misses the point. Reducing the correlation of biases can, at the margin, produce net gains in accuracy even if the group’s average competence is reduced. The lawyers’ superior competence in domains outside their expertise, even if it exists, can be overbalanced by the reduction in the correlation of bias across the group. Of course other professions have correlated in-group biases as well. But we are not comparing professions in the abstract; rather we are starting with a group of nine lawyers and asking what the effect would be of adding one or more nonlawyers. So long as the biases of the other profession, whatever it may be, are not perfectly correlated with those of lawyers, adding a member of that profession to a group of lawyers will reduce the correlation of bias in the group. The same points would hold, in mirrorimage fashion, for any nonlawyer professional group faced with a mixed docket of decisions, some of which required legal determinations; such a group would, on this logic, do well to add a lawyer. So the point is not to criticize lawyers; it is to suggest that professional diversity on the Court promises to maximize overall accuracy.
J C
143
Nonlegal Knowledge and Error Costs The upshot is this: Let us suppose that there are always right legal answers, but that sometimes the right legal answer just is the answer that obtains in some other domain of knowledge, specialized or unspecialized. We can say something general, even without knowing what the right answers are, about how a group of nine justices should be constituted if the social goal is to maximize accuracy, or to minimize the likelihood and gravity of mistakes. Under plausible conditions, I have argued, a Court with more than zero lay justices will make fewer mistakes in the subset of cases in which law looks outside texts and precedents—“legal” sources in conventional terms. This claim is clear enough where law looks outside itself to specialized domains of knowledge, because lay justices with relevant expertise can bring their expertise to bear, and because lay justices will do not worse than lawyer-justices in cases where the relevant nonlegal specialized knowledge is not the type that the lay justices possess. Even where law looks outside itself to nonspecialized knowledge shared by all members of the political community, the presence of lay justices can produce greater accuracy because lay justices lack the systematic biases of lawyers that arise from lawyers’ common professional training. More technically, the presence of lay justices reduces the correlation of biases across the group, increasing the effective number of independent votes and thus the group’s overall accuracy; indeed, this effect can obtain both in cases where law incorporates unspecialized knowledge about community values, and in cases where law incorporates specialized knowledge that does not lie in the lay justice’s area of expertise. Overall, lay justices should increase the Court’s accuracy in cases where law is not autarkic, and this is a benefit, all else equal.
Tradeoffs and Uncertainty So far I have suggested only that, under plausible conditions, there is an epistemic benefit to including lay justices on the Court. The magnitude of this benefit is uncertain, and I have not yet suggested that the benefit exceeds the epistemic costs. In this section, I will define the relevant costs and suggest that precisely because of the severe uncertainty about the magnitudes of the benefits and costs, opting for zero lay justices is a poor solution. The principal cost is that lay justices will, plausibly, increase the number of mistakes in what I have called autarkic cases—cases in which
144
L L R
law only looks to “legal” sources conventionally so-called. Nonlawyers will bungle the intricacies of doctrine; especially at the Supreme Court level, where many cases are hard in the sense that they are difficult, rather than indeterminate, strictly legal competence is at a premium. This argument from the technocratic expertise of lawyers has always been the main objection to lay justices. This tradeoff is a real one. Because it is real, we face an optimization problem. The problem is to find the number of lay justices, and the types of professional expertise those lay justices should have, that minimizes errors overall, netting out the reduced expected errors in cases in which law is not autarkic and the increased expected errors in cases in which law is autarkic. I have no direct information about the values on either side of the ledger; nor does anyone else, because we have never had even a single lay justice on the Court (in the sense described above). The current regime—zero lay justices—itself represents an uninformed guess about the best solution to the optimization problem across the total array of cases. And when a new seat comes open on the Court, there is no status quo set of qualifications for possible appointees; neither the appointment of a lawyer nor the appointment of a nonlawyer has any default priority or pretheoretical superiority.
Marginalism Under Uncertainty Despite the uncertainty, the optimal number of lay justices is quite likely greater than zero. This follows from the principle of marginalism: because there are diminishing marginal returns to expertise in group decisionmaking, the last increment of legal expertise is worth much less than the first increment, and the first increment of nonlegal expertise is worth much more than later increments. Suppose a Court of eight justices, with one vacancy, and a docket comprising both autarkic cases and nonautarkic cases. The ninth lawyer-justice will add a tiny extra increment of legal expertise, but the first nonlawyer justice will add much greater value to the Court in the domain of nonautarkic cases. This underscores an important practical point: the greatest marginal net benefit would be produced by appointing a first lay justice whose expertise is relevant to whatever is the most common subcategory of nonautarkic cases requiring specialized expertise. If, as some have suggested, the Court tends to do poorly in business cases and tax cases, then perhaps the first lay justice should be an eminent partner from an accounting or securities firm. Or, if it is a sensible prediction that biotechnology and
J C
145
medical ethics cases will take on an important share of the docket in the next generation, a doctor might be appointed. A very small number of lay justices—say, three—could diversify the Court’s knowledge and thereby cover a good portion of the professional landscape. A Court composed of six lawyers, one historian, one economist, and one soldier would rarely find itself deciding a case requiring expertise possessed by no one in the group. On the current Court, that situation is routine. Nothing in these points is unique to lawyers, as opposed to other professions. Exactly the same argument from marginalism would support adding a single lawyer to a panel of nine doctors charged with a portfolio of medical decisions, in some of which medical standards themselves incorporate legal issues. In general, under circumstances of grave uncertainty, it is a good rule of thumb for the institutional designer of a decisionmaking committee to diversify the committee along margins relevant to the decision, ensuring against the correlated biases and “groupthink”283 that can afflict a group of the likeminded;284 I return to this point shortly. One of the major sources of correlated bias is common professional training or background, or self-selection into a given profession by certain types. The many recent arguments for diversity along ideological and other dimensions make it all the more surprising that our highest court is so conspicuously undiversified, so monolithic, along the dimension of professional training. An important further question, at least in principle, is the number of cases in the autarkic and nonautarkic categories. If many cases are autarkic, the costs of lay justices are high; if there are many cases in which Supreme Court law is open to outside knowledge, then the benefits of lay justices are high. Here we are holding constant the composition of the docket and the content of the legal rules the Court announces. If lay justices affect the Court’s case selection and rulemaking, then more nonautarkic cases would be added, in turn increasing the benefits of having lay justices among the Court’s members. I return to these questions below. Casual scrutiny suggests that, surprisingly, many cases do incorporate outside knowledge, but there are no systematic studies. To believe that the optimal number of lay justices is zero, however, requires an implausible prior belief that almost all cases are autarkic. On marginalist grounds, the most sensible assumption is that adding the first increment of nonlegal expertise, and the first increment of negative correlation with lawyers’ 283 284
Irving L. Janis, Victims of Groupthink (1972). Cass R. Sunstein, Why Societies Need Dissent (2003).
146
L L R
professional biases, will bring net benefits, so long as some nontrivial fraction of cases run beyond autarkic law. This also assumes that the cost of errors that do occur are symmetric—that a mistake in a nonautarkic case is, on average, as harmful as a mistake is an autarkic case. For lack of any reason to think otherwise, symmetry of harms is the natural default assumption. Uncertainty, then, is not a problem for the argument; it is the cornerstone of the argument. We lack essential knowledge about the key variables: the competence of lawyer-justices and lay justices across various classes of cases, the degree to which lawyers’ biases are correlated, and the nature of the Court’s docket. It is impossible to prove, empirically, that the optimal number of lay justices is X rather than Y. But this means it is especially wrong to be complacent that the optimal number of lay justices is zero. Precisely because we are not sure what conditions obtain, it would be best to mix things up a bit, hedging our bets with at least a minimum of professional diversity. The solution we have adopted by unthinking default—zero lay justices—is too extreme. To be sure, even if we suspect that a prudent institutional designer concerned to minimize error would not choose the extreme solution, we do not know how many lay justices the Court should have: two? five? However, starting from zero, we might be able to move towards the optimum, and know that we are moving towards the optimum, even if we do not know exactly where the optimum lies. Why not experiment a bit, appointing one lay justice and watching how things go? I examine some of the costs and benefits of this sort of experimentation shortly.
Deliberation, Expertise, and Groupthink So far we have discussed mechanisms that aggregate the information held individually by members of the voting group. These mechanisms need not involve any sort of deliberation; as we have seen in Chapter 1, the logic of the Jury Theorem operates even if the voters do not talk to one another before voting. Voting without deliberation is not a misleading picture of the Court if, as some accounts suggest, the justices do little in the way of group deliberation even in conference before voting on cases.285 However, other Court insiders deny this, so let us suppose that there is real deliberation within the Court prior to voting. How is the argument for lay justices
285
Lawrence Blum, The Supreme Court 164 (6th ed. 1998).
J C
147
affected, if at all? What are the relationships among information, deliberation, and voting, within the Court and in general? In general, the relationships are complex, and the effects of deliberation on group performance are ambiguous, depending a great deal on the specifics of the case. It has been suggested that deliberation is inconsistent with the condition of the Jury Theorem that requires the aggregated votes to be independent of each other.286 However, as discussed in Chapter 1, this view is too simple. The Theorem requires statistical but not causal independence; independent votes cannot be predicted just by knowing how some other vote was cast, but voters can deliberate together and influence each others’ views without necessarily undermining the Theorem’s logic.287 Even if there is deliberation among the justices, the Jury Theorem may continue to operate. However, as Chapter 1 explained, deliberating groups can suffer pathologies that hamper group performance, including various forms of herding, informational or reputational cascades, and group polarization. In a range of settings, deliberation among the likeminded can make decisions worse, not better, than voting without deliberation. In other cases, however, decisionmaking groups do better by deliberating before voting, especially where a modicum of diversity—of information, judgments, and outlook—is present among the group’s membership. These ideas suggest that adding at least some lay justices might improve the Court’s decisionmaking. Two mechanisms are especially important here. Lay justices with relevant expertise can improve deliberation by making self-confirming contributions. And lay justices, with or without relevant expertise, can block a group of lawyer-justices from herding towards the wrong answer. In general, professional diversity dilutes groupthink and can thus improve both group deliberation and group decisionmaking. First, where one member of a group has genuine expertise in a relevant area, the expert may have outsized influence in steering the group towards correct answers. Experts will sometimes offer so-called “eureka” contributions288—arguments or information that are not obvious to the uninformed before the expert speaks, but that are self-confirming and 286 287 288
Bernard Grofman and Scott L. Feld, Democratic Theory and the Public Interest: Condorcet and Rousseau Revisited, 83 Am. Pol. Sci. Rev. 1317 (1989). See David M. Estlund, Democratic Authority: A Philosophical Framework (2007). See Garry Stasser and Beth Deitz-Uhler, Collective Choice, Judgment and Problem-Solving, in Blackwell Handbook of Social Psychology: Group Processes 31–55 (Michael A. Hogg and Scott Tindale eds., 2002); see also Cass R. Sunstein, Infotopia: How Many Minds Produce Knowledge 60–64 (2006).
148
L L R
patently true or sensible once uttered. More generally, in deliberating groups, sometimes experts will prevail simply because all concerned will recognize the force of the better argument. Where a lay justice has professional expertise relevant to the case at hand, even a dollop of that expertise can provide large benefits to the Court’s deliberation. Second, the presence of lay justices can block herding towards wrong answers by a group of lawyer-justices. Herding can arise even in groups all of whose members are fully rational; it arises because each individual rationally assumes that the views of others rest on information that together swamps the value of the individual’s own information.289 However, the presence of better-informed individuals, who tend to resist the group judgment when it contradicts their own, can shatter information cascades.290 This mechanism can work even where, indeed because, the nonconformist is excessively confident in her own judgment. Under plausible conditions, groups with some mix of overconfident members—members who put too much weight on their own information—actually do better than groups where all members form fully rational beliefs, because the overconfident members disclose more private information to be aggregated in the group decision.291 Where the lay justice has relevant information not held by lawyers, and even if the lay justice is unreasonably confident in her own information or judgment, cascades in bad directions can be broken. In such cases, the lay justice contributes just by taking a different perspective than the group of lawyer-justices. To be sure, herding can occur in good directions as well as bad, and the presence of dissenters in a decisionmaking group creates costs as well as benefits. In some cases the lay justice will resist lawyerly cascades in desirable directions—perhaps towards the right legal answer in cases where law is autarkic. Here too, however, the question is whether the marginal benefit of including at least one lay justice in a group of lawyers outweighs the marginal harm. The addition of the first lay justice produces the largest benefits in terms of outside expertise and dilution of likemindedness, while the marginal move from nine to eight lawyer-justices produces only a small reduction in the group’s overall legal expertise. 289 290 291
Sushil Bikhchandani et al., A Theory of Fads, Fashion, Custom, and Cultural Change as Informational Cascades, 100 J. Pol. Econ. 992 (1992). Sunstein, supra note 288, at 66–67. See Antonio E. Bernardo and Ivo Welch, On the Evolution of Overconfidence and Entrepreneurs, 10 J. Econ. & Mgmt. Strategy 301 (2001). The presence of overconfident members “can be useful if groups are large enough to benefit from the positive information externality [supplied by the irrationally overconfident members], if individuals have low-precision information, and if overconfidence is moderate rather than extreme.” Id. at 303.
J C
149
The point unifying these mechanisms is that many deliberative pathologies arise from groupthink, in one sense or another; and professional diversity dilutes groupthink. A group of likeminded justices may err in similar directions because of common professional training or selfselection into the profession. Above, we interpreted the effects of common training or common prejudices strictly in terms of correlated biases in a voting group; that interpretation does not assume any exchange of information or views within the group before voting occurs. Here we interpret the same point in deliberative terms. Where views are exchanged, a group containing professional diversity will do better, under plausible conditions, than a professionally homogenous group. Crucially, we need only consider these costs and benefits at the margin, not in the abstract. To hold the view that the optimal number of lay justices is zero, one must believe that the deliberative costs of subtracting the ninth and last lawyer-justice in cases where the law is autarkic outweigh the deliberative benefits of introducing the first lay justice in cases where the law incorporates outside knowledge. While this is possible, it is unlikely, so long as there are diminishing marginal benefits to legal expertise.
Dual Competence? Even if the foregoing argument fails, there is a weaker version that may succeed: the Court should have more justices of dual competence than at present. There are now programs at major law schools that provide joint degrees in law and medicine, or law and business. Nor are formal degrees necessary; a well-motivated president might appoint Richard Posner, a lawyer with no economics degree who has made major contributions to economics. In the same vein, commenting on the idea that lay justices with political experience might be beneficial, Christopher Eisgruber writes: I am inclined to believe that we might get equal benefits from less radical changes to the composition of the Court, such as by selecting lawyers with political experience to fill vacancies. A president inclined to nominate Bill Bradley could instead call upon Bruce Babbitt or George Mitchell.292 292
Christopher L. Eisgruber, Constitutional Self-Government and Judicial Review: A Reply to Five Critics, 37 U.S.F.L. Rev. 115, 157 (2002). See also Akiba Covitz and Mark Tushnet, Political Appointee, The New Republic Online, Sept. 29, 2005, available at https://ssl.tnr.com/p/ docsub.mhtml?i=w050926&s=covitztushnet092905.
150
L L R
I believe that Eisgruber’s point generalizes well beyond politics, even if politics is conceived as a professional career that requires objective expertise. The status quo composition of the Court is extreme along more than one dimension: not only does it contain all lawyers, it is also the case that all the Court’s lawyers are lawyers alone. We could have more politician-cum-lawyer justices, but also more economist-cum-lawyer justices, historian-cum-lawyer justices, and so on. I offer the dual-competence argument as a fallback claim, but we should underscore how large a change it would effect compared to the current baseline. On the current Court none of the justices is dual competent in any strong way, at least if one puts aside Justice Stephen Breyer, an expert in regulatory policy and economics. In the public sphere, debates over judicial appointments would look far different if genuine expertise in some other body of knowledge or skill, along with sterling legal credentials, were seen as a necessary qualification. If the dual-competence argument is correct, we would not have a full argument for lay justices, but we would have the best half of the loaf—an argument against the current domination of the Court by pure lawyer-justices. That said, however, the dual competence argument poses, without answering, a crucial conceptual question: even if it is desirable to have dual competence, should dual competence be built in at the level of individuals or at the aggregate level of the group? The group of justices that contains both pure lawyers and pure nonlawyers is dual competent as a group even if none of its members is dual competent, taken one by one. Groups may have emergent properties by virtue of aggregation; it is not clear why dual competence at the level of the group should be thought inadequate. It is tempting but wrong to think that dual competence offers a free lunch—that we can have accountants for the nonautarkic accounting cases who are also lawyers for the autarkic cases, and so on, without costs on other margins. But there are such costs, precisely because the dual-competent justice also has legal training. In the categories developed above, the Court’s docket consists of (1) autarkic cases and (2) nonautarkic cases, with the latter set divided into two subsets: (2a) cases that draw upon specialized nonlegal knowledge and (2b) cases that draw upon nonspecialized knowledge of social values. (2a) further divides into (2aa) cases in which the nonlegal expertise at issue is relevant (economics in a financial case) and (2ab) cases in which it is not relevant (medicine in a financial case). We must then compare the performance of two groups: a Court composed of eight pure lawyers and one economist, for example, with a Court
J C
151
composed of eight pure lawyers and one dual-competent lawyer-economist. (2aa) is a wash; either the pure lay justice or the dual-competent justice will improve the group’s performance where relevant nonlegal expertise is needed. As compared to appointing a purely lay justice, dual competence improves the average competence of the group in (1). But because the dual-competent justice is a lawyer with the common training and professional blind spots of the lawyer, the presence of the dual-competent justice increases the correlation of biases across the group, and thus reduces the group’s performance at least in (2b) and perhaps also in (2ab). We need to consider possible crosscutting effects of training, but these are ambiguous. The dual-competent justice’s training in the other field may dilute his lawyerly biases, ameliorating the costs in (2b) and (2ab). But it is not clear that lawyerly skill is unaffected when lawyerly biases are diluted, so the benefits of the dual-competent justice may be lower in (1); moreover, the dual-competent justice’s training as a lawyer may just as well dilute his expertise in the other field, in which case the pure lay justice may do better than the dual-competent justice in (2aa). It is unclear how these costs and benefits net out. But three conclusions are possible. First, even those who are convinced that legal training is an indispensable prerequisite for all nine of the Court’s seats might still agree that we should have more dual-competent justices than at present, perhaps many more. Second, it is an unsupported prejudice to assume that a dual competent-justice must necessarily bring greater net benefits to the group’s overall performance than would a pure lay justice; all depends on factors such as the strength of the common biases induced by lawyerly training or self-selection, and the risks and harms of errors in various subsets of autarkic or nonautarkic cases. Third, lay justices and dual competence are not mutually exclusive ideas. Even if as many as seven of the Court’s nine slots are permanently reserved for pure lawyers, we might appoint an eighth dual-competent justice and a ninth lay justice, and see how things work out.
The Representation of Expertise So far I have provided an epistemic argument for nonlawyer justices, or at a minimum, for dual competence. However, when law looks outside itself to other domains of knowledge, there might be other means by which a group exclusively composed of pure lawyer-justices could improve their decisionmaking in cases in which law is not autarkic—means that might produce greater overall benefits than putting lay justices on the Court.
152
L L R
These include litigants’ arguments, amicus briefs, deference to administrative agencies or court-appointed experts, and so forth. The problem is somewhat analogous to a firm’s problem of “makeor-buy,” a standard issue in economic theory.293 The analogy is slightly inexact, because law clerks and other judicial staff, unlike the parties’ lawyers, agencies, and special masters, are employed inside the judicial firm rather than consulted ad hoc, so are on the “make” side rather than the “buy” side, but the parallel may still be useful. Should the judicial firm, the Court, employ its own experts to manufacture expertise from within, or should it buy expertise case-by-case? I suggest that the crucial epistemic advantage of having at least some lay justices on the Court is the representation of expertise. Having nonlegal experts participate in the selection of cases for full hearing, and in deliberation, provides an institutional foundation for nonlegal expertise, one that cannot be replicated through case-by-case mechanisms for incorporating expert views. The latter mechanisms will be stymied by the epistemic bottleneck that arises when only lawyers decide whether and how to incorporate nonlegal expertise.
Make or Buy? A bench solely composed of pure lawyer-justices might use many techniques to incorporate extrinsic expertise into its decisions. Among these are the following (doubtless others might be added):
Deference to Administrative Agencies Under the Chevron rule,294 lawyer-justices defer to the reasonable views of administrative agencies. Chevron grants agencies a “policy space” in which to make technocratic and democratic judgments, rather than having to provide a “point estimate” of the legal answer that they think will find approval with reviewing judges.295 Lawyer-justices themselves benefit from providing agencies with this policy space, insofar as doing so encourages agencies to base decisions on their expertise, because lawyer-justices can incorporate the resulting expertise into their own decisions. Beyond agencies properly so-called, judges draw upon the views and decisions of 293 294 295
Oliver E. Williamson, Market and Hierarchies: Analysis and Antitrust Implications (1975). Chevron v. NRDC, 458 U.S. 1267 (1984). E. Donald Elliott, Chevron Matters: How the Chevron Doctrine Redefined the Roles of Congress, Courts, and Agencies in Environmental Law, 16 Vill. Envtl. L. J. 1 (2005).
J C
153
a bewildering variety of more or less specialized nonjudicial bodies and tribunals, such as Article I courts and advisory commissions, with the level and nature of judicial deference varying according to the case.
Litigants, Expert Witnesses, and Amicus Briefs Litigants will attempt to import expertise when it benefits them to do so, through expert witnesses at trial and selective citation of authorities on appeal. Looking beyond the litigants themselves, courts routinely permit the Department of Justice and the Solicitor General, other federal agencies, interest groups, advocacy groups, coalitions of academics, and other nonparties to file amicus briefs with the court. These briefs are sometimes tendentious but can provide valuable factual information or serve as informative signals of the views and judgments of outsiders. With regard to the central holding of United States v. Virginia, that VMI’s adversative training method would not be undermined if women were admitted, it is sensible to ask “[h]ow does the Court know?”296 Part of the answer, if there is one, must involve the wide range of amicus briefs presented to the Court in that case.
Special Masters and Law Clerks Between soliciting the views of outside experts, on the one hand, and appointing experts to the Court, on the other, there is a middle ground: experts might be employed as judicial adjuncts, either for particular cases, as is the practice with special masters, or as more permanent personnel. One might imagine a justice having a preference for hiring law clerks with joint degrees or other nonlegal expertise; it is unclear whether any current justice has such a preference, but as joint degrees become more common it would not be shocking to see it develop. Are these and other mechanisms for incorporating expertise superior to putting nonlawyers, expert in other fields, on the Court? What we are interested in, of course, is really the optimal mix of mechanisms. A Court with more than zero nonlawyers or dual-competent lawyers would surely continue to use amicus briefs, and would be right to do so. The question then becomes whether mechanisms for incorporating external expertise into the decisions of a Court of nine pure lawyers can provide a full substitute for actually appointing a nonlawyer or a dual competent lawyer to the Court—whether the external mechanisms should be the exclusive means for generating expertise. This is akin to the make-or-buy decision that faces firms in ordinary markets. Were transaction costs zero, all 296
Richard A. Posner, Against Constitutional Theory, 73 N.Y.U. L. Rev. 1, 13 (1998).
154
L L R
productive resources could be assembled on spot markets for particular transactions, and no firms would exist.297 Likewise, were there zero transaction costs to incorporating expertise through extrinsic mechanisms, no internal judicial expertise would be necessary. For two reasons, however, these considerations do not require rejecting the argument for lay justices or dual-competent justices. First, the argument from incorporation of external expertise is double-edged, because nonlawyer justices could also incorporate legal expertise from outside sources. Indeed, in a world of zero or very low transaction costs, a Court could be composed solely of pure nonlawyers and could then incorporate legal expertise as needed. Second, in our world of positive transaction costs, there are benefits from having nonlawyers or dual-competent lawyers on the Court that cannot be replicated through case-by-case mechanisms. I will treat these two points in turn.
Incorporating Legal Expertise Expertise can be incorporated through mechanisms of the sort we have canvassed. But nothing about that point necessarily favors having a Court composed of pure lawyers. The point applies to all expertise, including legal expertise, which could also be incorporated as needed. A lay justice or justices, for example, might rely on law clerks, or on amicus briefs, or on law review articles to help clarify the legal issues in particular cases. At the extreme—in a world of zero transaction costs—we could have a Court composed solely of pure nonlawyers, who incorporate legal advice from lawyers on a case-by-case basis. It would not be helpful to claim that, where legal expertise is incorporated, the lawyers who do the advising would be acting as the “real” justices, or that nonlawyer-justices will misunderstand the external legal sources. Symmetrically, one might equally say that, in cases where the justices take the advice of nonlawyers on subjects that the law makes decisive, the “real” decisionmakers are the nonlawyers, or that lawyer-justices will misunderstand the nonlegal sources. Of course transaction costs are not zero; across the array of cases the Court actually hears, the most efficient course is surely to have some lawyers as permanent members of the Court. But this emphasizes that the presence of transaction costs cuts in both directions. If there are advantages to incorporating lawyers directly into the Court’s composition, there 297
Ronald H. Coase, The Nature of the Firm, 4 Economica 386 (1937).
J C
155
are symmetrical advantages to incorporating nonlawyers as well. Suppose that some members of the Court should be lawyers because nonlawyers cannot fully evaluate the arguments of lawyers, or because legal considerations will be slighted unless a trained lawyer in present to argue for them, or simply because it is cheaper to have a legal expert permanently on staff than to seek out an expert anew in every case. The same might just as well be said about other types of expertise. In a world of positive transaction costs, there are institutional benefits to bringing desired forms of expertise within the institution’s walls. And this is true of both legal and nonlegal expertise.
The Benefits of Representation What, concretely, are the costs and benefits of incorporating expertise from external sources or of “producing” expertise internally? Here I want to suggest that there are important benefits to building desired forms of nonlegal expertise into the very personnel and composition of the Court. We may call these “the benefits of representation”—where what is being represented are various bodies of professional knowledge and skill, rather than geographic districts of voters or any of the other represented entities familiar from democratic politics. The leading argument for ideological diversity on the Court gives two reasons for building that form of diversity into the Court’s composition. The first involves agenda-setting: the Court’s docket is essentially discretionary, and “[j]udges with distinctive views notice legal problems that other judges do not see—not through ignorance or malice, but because of differing priorities.” The second involves deliberation: “if they are willing to listen, judges of one basic outlook will learn a great deal from those with other basic orientations. . . . [D]ifferences in perspective often improve both the collective reasoning process and the outcomes.” By contrast with these two advantages, the alternative—to incorporate ideological diversity through the arguments of litigants—is “inevitably inadequate,” because “even the best advocate usually plays only a limited role in comparison with a member of the Court. Divergent views should be presented, and pressed, during internal deliberations, when the Court is formulating results and reasons.”298 This line of argument says, in essence, that differing views must be incorporated into the Court’s decisionmaking process through structural 298
Strauss and Sunstein, supra note 241, at 1150–55.
156
L L R
representation—that there is no close substitute for having people of different views actually in the room where decisions are made. The same is true for differing bodies of professional expertise and for professional diversity. At the agenda-setting stage, prominent voices have complained that the Court does not take enough cases involving tax, accounting, and business, in part because as lawyers the justices’ interests mostly do not run in that direction.299 The same complaint might be made about other areas, such as complex environmental or antitrust or energy cases. At the stage of argument, deliberation, and voting on the merits, it is plausible that the mechanisms we have surveyed—Chevron deference, briefs by litigants and amici, joint-degree law clerks, and so on—do not provide adequate substitutes for having lay justices on the Court. Those sources of external expertise—which will often conflict, or be tendentious or misleading, or just confusing—must ultimately be evaluated by the members of the Court, whose ultimate deliberations occur, to the extent they occur at all, in a secret conclave. There is good reason to worry that, if only lawyers are doing the evaluating, important information or ideas will never be heard or will be mishandled, where an expert evaluator would not have made those mistakes. In general, structural representation is indispensable where there is a gulf between having A attempt to take account of B’s views, on the one hand, and having B herself present those views and participate in the decision, on the other—even if A acts in the best of faith. Why might such a gulf arise? Again, the arguments for ideological, racial and gender diversity implicitly suggest several plausible mechanisms. One is that information costs and cognitive limitations will prevent A from even conceiving of the arguments or ideas B would offer, whereas B might present those arguments quite easily if sitting in the room. Another is that disparities in background or training or experience will prevent A from appreciating the force of B’s arguments; even if they occur to A in a pallid form, they will not pack sufficient emotional punch to influence A’s deliberation or voting, whereas face-to-face exchange with B, and the need to accommodate B as an equal member of the voting group, will bring home the force of B’s arguments. These problems are examples of the problem of epistemic bottlenecks, discussed in Chapter 1; structural representation is indispensable to eliminating the bottleneck.
299
Cf. Kenneth W. Starr, The Supreme Court and Its Shrinking Docket: The Ghost of William Howard Taft, 90 Minn. L. Rev. 1363, 1366, 1384–85 (2006).
J C
157
Problems and Implications I have presented the main line of the argument and examined a major institutional alternative. Here I will touch upon a cluster of remaining issues and tie up some loose ends. I assume, in each case, that the relevant considerations pro and con apply to dual competent-justices, but in a weakened or diluted form on both sides of the ledger, so I will not explicitly discuss them as a separate category.
Endogenous Agendas, Endogenous Rules? Another simplifying assumption made above was that the Court’s agenda and the legal doctrines it declares are exogenous to the Court’s professional composition. Perhaps a Court that appointed a lay justice expert in profession P would find itself hearing many more cases implicating the concerns of P in one way or another—perhaps because the P-competent justice would contribute the decisive fourth vote to take those cases. Perhaps the legal rules the Court would declare on the merits would change their character as well, more often incorporating, as the right legal answer, knowledge from the relevant domain. It is even possible that a runaway scenario would result: the Court might hear many more cases in which law is not autarkic and develop many more nonautarkic rules, which would increase the demand for lay justices, who would vote to hear more such cases, and so on, to what limits no one can say. What would be so bad about this? The idea must be that if Courtdeclared constitutional and statutory law became too dependent upon knowledge from other domains, too wide open to breezes from other disciplines and other professions, its distinctive character would be lost. It is hard to evaluate this view because it is not clear what theory or theories could help us decide what the docket should look like, or what subset of Court-declared rules should incorporate outside knowledge by reference. Law is a “partially autonomous discipline,”300 influenced and penetrated by other domains of knowledge but not wholly reducible to them, and that remains so whether many cases in the Supreme Court are autarkic or not. Even if the proportion of nonautarkic cases rose dramatically at the Supreme Court level, this might reflect nothing more than an optimal division of labor between the Court, on the one hand, and lower courts, 300
Richard Fallon, A Constructivist Coherence Theory of Constitutional Interpretation, 100 Harv. L. Rev. 1189, 1264–65 (1987).
158
L L R
on the other. Perhaps the Court spends far too much of its time now on cases in which law is autarkic—on fiddling with the impenetrably selfcentered details of law. Indeed, if the Court’s agenda and rulemaking are endogenous, one worry about lay justices that might otherwise seem plausible becomes less so. The main costs of lay justices are incurred in cases where Supreme Court law is autarkic, while the main benefits arise in nonautarkic cases. If adding lay justices to the Court reduces the proportion of autarkic cases or rules, then the costs of lay justices go down and the benefits of lay justices go up. In this scenario, appointing lay justices would be a kind of self-fulfilling or self-confirming good. Now let us suppose that, for some reason, the runaway scenario would be bad, not good. Still, it seems unlikely to be worth worrying about. The Court’s agenda and its scope for choosing legal rules are not as elastic as all that. The Court inevitably takes a substantial proportion of cases and announces a substantial proportion of legal rules that are not autarkic, despite the current and historic total dominance of lawyers on the Court. The demands of operating a legal system that regulates many nonlegal domains constrains the bench of lawyers to take a substantial fraction of cases in which law is open to outside knowledge. A symmetrical constraint operates in the other direction. Even a Court composed solely of nonlawyers would surely be constrained to hear many cases, perhaps a majority of the docket, in which law is utterly autarkic. Such cases are the meat-and-potatoes of any legal system and on simple statistical grounds make up the bulk of occasions on which circuit courts disagree, or issues of unusual public importance are raised, or in some other way the Court’s criteria for hearing cases are implicated. Overall, the runaway scenario is unlikely to occur; nor is it clear that a modest increase in the number of nonautarkic cases on the court’s docket would be bad.
Intracourt Deference The argument assumed as well that each justice votes independently on all cases, in the sense that no justice defers to the views of any other. If that assumption is relaxed, perhaps there is a problem. Within a voting group, deference by some to others reduces the number of effectively independent votes and can thus reduce the group’s aggregate accuracy. Perhaps the problem is that lawyer-justices will excessively defer to lay justices where law incorporates specialized nonlegal knowledge, making the lay justice(s) in effect the sole decisionmaker(s).
J C
159
However, this argument goes much too quickly. It is wrong to compare (1) a Court composed of nine lawyer-justices in which all vote independently, so that no intragroup deference occurs, with (2) a Court with more than zero lay justices in which intragroup deference does occur. This is a false comparison because intragroup deference occurs and historically has occurred among the lawyer-justices as well. Justice Harry Blackmun was widely reputed to be “the tax justice,” to whom colleagues deferred on the complex questions of accounting, finance, and statutory interpretation presented by tax cases. If other justices are going to defer to someone on tax questions in any event, it might well be an improvement to have them deferring to a professional tax accountant rather than a lawyer-justice who—however successful or unsuccessful his tax opinions301—was basically moonlighting. Deference occurs under either regime, so raising the competence of the justice who is afforded deference improves matters. It is even unclear, in the first place, that intragroup deference reduces group accuracy; it might or might not do so, depending upon the facts. Under the Jury Theorem, group accuracy is a function of (1) the number of independent voters and (2) the average competence of voters. Where low-competence voters defer to high-competence voters, the number of independent votes is effectively reduced, but the average competence of the voting group goes up; whether the gain is greater than the loss depends entirely on the precise values of these variables.302 A small number of (effectively independent) voters of very high average competence can do better than a large number of voters of lower average competence. So it is hard to know, even in principle, whether deference by lawyer-justices to lay justices on matters within the lay justice’s specialty would increase or decrease the Court’s overall accuracy. The converse concern is that lay justices will excessively defer to their lawyer-justice colleagues. In the broader debates I adverted to above, opponents of lay judges hope for deference by lay judges to their legallytrained colleagues, so it is unclear that anyone will actually see this as a concern. Still, several brief points are useful. As above, under some conditions intragroup deference can actually increase the group’s overall accuracy, and in any event harmful deference can occur even within a group of nine lawyer-justices. Moreover, one mechanism that is often said to
301 302
Robert A. Green, Justice Blackmun’s Federal Tax Jurisprudence, 26 Hastings Const. L.Q. 109 (1998). See Estlund, supra note 287.
160
L L R
produce deference by lay adjudicators to legally-trained adjudicators is that the lay adjudicators are at best part-timers, and are sometimes even single-shot participants in the system, as are Anglo-American jurors. Players of that sort will often be dominated by professional repeat players, who know the institution’s rules and norms. This sort of mechanism does not apply to nonlawyer justices, who will be repeat players in the work of the Court as much as their legally-trained colleagues.
Philosopher-Kings and Accountant-Justices A standard complaint about the Supreme Court, especially where constitutional judicial review is concerned, is that the justices are like “philosopher-kings.” An implication of the argument here is that we might take this point more literally, and more seriously, than those who make it intend. If law generally, or constitutional law in particular, incorporates specialized knowledge to which philosophers have privileged access, perhaps we should have a professional philosopher on the bench. Whether that is so depends on whether philosophy generates usable objective knowledge relevant to law—a much-debated question.303 Whatever the case with philosophy, there are undoubtedly disciplines that do generate specialized bodies of objective and usable knowledge relevant to law. I have mentioned several such disciplines, but only by way of example. If the examples are bad ones, no matter; whatever disciplines turn out to make the cutoff, the important thing is to ensure that at least some of the disciplines that do make the cutoff have representation on the Court. A corollary is that pseudo-disciplines that do not meet these conditions do not fall within the proposal. Although appointing an astrologer to the Court might dilute lawyerly biases, the astrologer would have nothing else to offer and would thus be strictly inferior to a professional with genuine expertise in any class of cases that reach the Court. An appointee of that nature would supply both increased diversity and relevant competence. Suppose that economics is one of the disciplines that make the grade. We might appoint a professional economist to the Court; perhaps 303
On the question whether moral philosophy is objective knowledge, see Ronald Dworkin, Objectivity and Truth: You’d Better Believe It, 25 Phil. & Pub. Affairs 87 (1996); on the question whether it is usable objective knowledge, given intractable disagreement, see Jeremy Waldron, The Irrelevance of Moral Objectivity, in Natural Law Theory: Contemporary Essays (Robert George ed., 1992); on the question whether it is relevant to law, compare Ronald Dworkin, In Praise of Theory, 29 Ariz. St. L.J. 253 (1997) with Cass R. Sunstein, Response: From Theory to Practice, 29 Ariz. St. L.J. 389 (1997).
J C
161
Justice Andrei Shleifer, an economist with a strong interest in law, would add a great deal of economic sense to the Court’s decisions while detracting only a little, if at all, from its aggregate doctrinal skills. If philosophy and economics are too rarefied, more mundane suggestions may seem more appealing. We might do well to appoint a trained accountant, historian, doctor, environmental scientist, or military officer to the Court. At a minimum, dual-competent figures come to mind as potential justices. I do not know how many lay justices or dual-competent justices there should be, or what the priority of appointment should be across nonlegal disciplines; much would depend on facts, about which types of nonlegal specialized knowledge are most often incorporated into law, and which types of professional would add most value to the Court’s decisionmaking. On the present state of empirical knowledge we can do little better than guess at these things. What I have emphasized, however, is that our current practices already and inevitably embody a guess about the same questions—the content of that guess being that zero lay justices is the best solution. There is no reason to think that is so, and there is no natural or default priority to the practice of putting only pure lawyers on the Court. Even if we do not know what the right number of lay justices is, we can be as confident as the deep uncertainty will permit that by appointing at least one, we will at least be moving towards the optimum.
Interactions How does the claim I make in this chapter interact with the proposal, in chapters 2 and 3, to shift a great deal of authority to develop common-law constitutionalism from courts to legislatures? The two claims are substitutes. Legislatures’ superior epistemic capacities stem from their relative greater diversity, which often trumps competence. Increasing the Court’s epistemic diversity would then dilute the epistemic superiority of legislation. However, the existence of this substitution relationship does not entail that we must choose between the two claims altogether. The epistemic optimum might require that a balance be struck, in which some constitutional authority is transferred to the legislature while some epistemic diversity is introduced into the Court. Moreover, if either claim is held constant, the other applies with undiminished force. If the proposal offered in chapters 2 and 3 fails, then the proposal in this chapter becomes all the stronger, and vice-versa.
162
L L R
Conclusion: Lay Justices and Epistemic Legalism Bentham’s critique of the common law spilled over into a critique of the legal profession’s sociology, in part because the common law did not develop or exist in a sociological vacuum. So too, a deep assumption of epistemic legalism is that the judges will all be lawyers, even in our own system, in which common-law methods are applied more or less openly to the interpretation of written texts, such as statutes and the Constitution. The epistemic critique of epistemic legalism—my negative thesis, to the effect that epistemic capacities often imply a smaller rather a larger role for judicial lawmaking—must therefore be given a sociological foundation. I have attempted to update and fortify a version of Bentham’s claims, a version that casts those claims in epistemic terms rather than in terms that impeach lawyers’ motivations. I suggest that even on strictly epistemic grounds, a modicum of professional diversity on the Court would better conduce to getting things right. Nonlawyers are indispensable for epistemic reasons, even in the very citadel of law’s empire.
Chapter 5
Unintended Consequences and Constitutional Amendments From Many Minds to Unintended Consequences The two main pillars of epistemic legalism, I have suggested, are manyminds arguments and arguments from the fear of unintended consequences. In this chapter I will focus on the choice between two alternative means of effecting constitutional change: formal amendment and common-law judicial interpretation. This inquiry amounts to a case study of the argument from unintended consequences. A standard view in American constitutional theory is that amendments are presumptively harmful, because the limits of reason that afflict their enactors guarantee bad and unintended results. Rejecting this view, I attempt to impartially examine the epistemic costs and benefits of common-law constitutionalism administered by judges, on the one hand, and constitutional amendment, on the other, as alternative means for updating the Constitution in the face of changing circumstances. I shall also briefly consider the claim that amendments are systematically futile, whether or not desirable. 163
164
L L R
The Problem, and a Standard Answer Constitutions obsolesce rapidly and must be updated over time to reflect changes in the polity’s circumstances and citizens’ values. What institution or process should be entrusted with the authority to do the updating? If periodic wholesale replacement of the Constitution is infeasible, the plausible choices are the constitutional amendment process, flexible interpretation by judges under the banner of constitutional common law, constitutional interpretation by legislatures through statutory enactment in a Thayerian regime, or some mix of these. I put aside statutes in this chapter and instead consider the choice between amendments and common-law constitutionalism by judges. (In the rest of the chapter, I will omit the qualifier “by judges,” which should be understood as implicit). At the end of the chapter, I will consider interactions between statutes and formal amendments. My main critical point is that the epistemic argument against constitutional amendment, rooted in a fear of unintended consequences, is hopelessly flawed. It rests on a nirvana fallacy that implicitly contrasts a jaundiced view of the amendment process with a romanticized view of constitutional common law, and that contrasts a romanticized view of the epistemic capacities of judges with a jaundiced view of the epistemic capacities of the legislators and other actors who pass amendments. Flexible judicial interpretation that updates the Constitution over time risks bad unintended consequences at least as much, and for many of the same reasons, as constitutional amendment.
The Question of Comparative Epistemic Competence Once we have dispelled this nirvana fallacy, constitutional updating can be seen to pose a question of comparative epistemic competence. Constitutional amendment, on the one hand, and constitutional common law, on the other, are epistemic alternatives for managing the inevitable updating of constitutional law over time. Under what circumstances might one process or the other prove epistemically superior, in light of the limits of foresight that afflict legislators, other participants in the amendment process, and judges? What institutional considerations, or variables, determine their relative performance? Relative to common-law constitutionalism, I will suggest, the amendment process is less focused on the facts of particular cases, puts less weight on the views of past judges, and allows for the participation of
U C C A
165
decisionmakers from a broader and more diverse range of professions and backgrounds. By identifying these variables, I hope to identify the empirical conditions under which, or the domain in which, either process is most likely to produce epistemically valuable constitutional change.
Constraints on Amendment: Structural and Intellectual I will begin by distinguishing the structural constraints on the amendment process from the intellectual constraints. The structural constraints are both legal and political. Legally, the U.S. Constitution is one of the most difficult in the world to amend formally; its stringent supermajority requirement, set out in Article V, ensures a lower rate of constitutional amendment at the federal level than in most states and most other constitutional democracies. Politically, sheer inertia and the desire to avoid formalized constitutional conflict or showdowns, which a series of straight-up votes on constitutional amendments produce, both limit the number of amendments that political actors will seriously attempt in the first place. The consequence of these structural constraints is an important substitution effect. Because formal constitutional change is difficult to secure through the Article V process, some constitutional change is shunted into the process of judicial interpretation and reinterpretation of constitutional text—into common-law constitutionalism. This substitution effect is real304 and inevitable, given demanding supermajority requirements for amendment. Formal amendment, in our current constitutional system, can never fully replace common-law constitutionalism, nor do I claim it should do so. However, that the structural constraints are real does not entail that the very low rate of amendment we currently observe is the highest rate that can realistically be attained. In past eras of American history, amendments have been used more frequently to effect needed constitutional change. In recent decades, however, the amendment process has seemingly atrophied,305 despite the absence of any changes to Article V (I ignore the bizarre case of the Twenty-seventh Amendment, proposed in 1789 yet not ratified until 1992). Part of what accounts for this, I suggest, is the growth of intellectual constraints on the amendment process. Legal elites have, by and large, developed an erroneous view that constitutional
304 305
See Donald S. Lutz, Toward a Theory of Constitutional Amendment, 88 Am. Pol. Sci. Rev. 355, 358 (1994). See Bruce Ackerman, The Living Constitution, 120 Harv. L. Rev. 1737 (2007).
166
L L R
amendment is suspect, in important part because of the risk of bad unintended consequences. That view is what I mean to challenge. If it can be dispelled, then one constraint at least can be loosened. Although at some point an increasing rate of amendments will bump up against the structural constraints, that point will be appreciably higher than the rate of amendment that can currently be observed, which is near zero.
Are Amendments Futile? One can further subdivide the argument from unintended consequences into two types: an argument that amendments will, against their enactors’ intentions, turn out to be futile, and an argument that they will, against their enactors’ intentions, turn out to be bad. Before turning to the claim that amendments are epistemically undesirable, let us briefly consider the claim that they are futile, whether or not desirable. University of Chicago law professor David Strauss argues, with impressive historical command, that amendments are systematically irrelevant, by which he means that constitutional law would look the same without them. With some qualifications, Strauss holds that “our system would look the same today if Article V of the Constitution had never been adopted and the Constitution contained no provision for formal amendment. . . .”306 As against this striking claim, I suggest that even if constitutional amendments are neither necessary nor sufficient to produce legal change, they may nonetheless be causally efficacious in producing legal change. Strauss’ irrelevance thesis overlooks that the causal force of amendments is probabilistic: even if an amendment neither guarantees a desired legal change nor is indispensable to it, the amendment may nonetheless make the change more likely than it would have been in the amendment’s absence. Some cause C that is neither necessary nor sufficient for producing some effect E may nonetheless make E more likely, perhaps far more likely, than E would have been in C’s absence. Having Babe Ruth on the team was neither necessary nor sufficient for the 1927 Yankees to win the World Series—they might have won anyway, and they might not have won even with Ruth—but it certainly helped. The same point holds as against the idea that amendments are systematically futile or irrelevant because amendment drafters cannot foresee all later questions and cannot control all the decisions of later interpreters. 306
David A. Strauss, The Irrelevance of Constitutional Amendments, 114 Harv. L. Rev. 1457, 1459 (2001).
U C C A
167
Of course initial drafters cannot settle all relevant questions; of course subsequent enforcement is partly, but only partly, at the mercy of subsequent generations; of course hostile judges may interpret the amendment grudgingly. But these possibilities do not make amendments meaningless. Rather, passage of an amendment systematically tends to push later outcomes in the directions desired by the amendment’s proponents, although as a probabilistic mechanism it provides no guarantee. Thus Strauss acknowledges in many cases, explicitly or implicitly, that some particular amendment actually brought about some relevant legal change. But he then argues for irrelevance by appealing from facts to counterfacts. Strauss suggests, for example, that the Thirteenth Amendment’s only practical effect was to abolish slavery in the border states. Even without the Amendment, he adds, “in any event, Congress very likely would have outlawed [slavery], and the Supreme Court might have upheld Congress’ action.”307 One wonders why this sort of appeal from facts to counterfacts should be permitted, however. A criminal defendant cannot excuse his acknowledged deed of homicide by showing that the victim would soon have died from other causes. No causal sense of irrelevance can be established by showing that some other legal instrument or institutions would have produced the effect that was, in fact, produced by a constitutional amendment. Strauss rightly showcases one excellent example, the Reconstruction Amendments, in which subsequent generations of judges and other officials nullified amendments for a time through grudging interpretation and sheer defiance. But even in that example, the amendments lay around waiting to be picked up and used by yet later generations, as they actually were, especially during the Warren Court, and most other amendments have been implemented straightforwardly. There is no reason to think that the Reconstruction example is broadly representative of constitutional amendments in general. Strauss is entirely correct that an actually operative cause can be irrelevant in a counterfactual sense. The airline that actually takes us from Point A to Point B might be dismissed as counterfactually irrelevant, in the sense that we could have gone by rail instead. I do not deny that “irrelevance” can be given this sense, but I do deny that this strictly counterfactual sense of irrelevance holds much interest in practical affairs. Consider that, had we gone from A to B by train instead of plane, we could argue in turn that the railway was counterfactually irrelevant, on exactly the 307
Id. at 1480.
168
L L R
same grounds. Whenever there are alternative means for accomplishing an end, all such means will be irrelevant in the counterfactual sense. It would, however, be very odd to conclude that it is immaterial which means is actually used. Counterfactual irrelevance does not at all entail that the alternative means are equally costly, or legitimate, or equivalent in any other respect. So even if our constitutional rules would be the same today had the amendment process laid out in Article V never existed, it might also be true that we are much better off because Article V did exist—if, for example, the Article V process produced the same set of constitutional rules, but at lower social cost than judicial updating would have. To know whether Article V makes us better off, we would have to compare the benefits and costs of formal amendments and judicial updating as alternative means of constitutional change. I undertake that inquiry next, although only as to epistemic factors. Here my point is that counterfactual irrelevance provides no help with the crucial comparative questions. In the end, it is unclear that counterfactual irrelevance is a property anyone should care about. Bentham would have dismissed it as a tool of common-law obscurantis.
The Epistemic Argument Against Constitutional Amendment Let us turn to the claim that amendments are undesirable by virtue of the limits of reason. Consider the following passage, which encapsulates a widespread view among legal theorists: [R]ecent Congresses have been stricken with amendment fever. More constitutional amendment proposals have been taken seriously now than at any other recent time. Some have even come close to passing. . . . Many of these amendments are bad ideas. But they are dangerous apart from their individual merits. [The Constitution] should be amended sparingly, not used as a chip in short-run political games. This was clearly the view of the framers, who made the Constitution extraordinarily difficult to amend.308
308
Kathleen M. Sullivan, What’s Wrong with Constitutional Amendments, in Great and Extraordinary Occasions: Developing Guidelines for Constitutional Change 39 (1999) (emphasis added) [hereinafter Sullivan, Constitutional Amendments].
U C C A
169
According to this view, Americans have traditionally been reluctant to amend the Constitution, for good reasons. Prominent among these reasons is an epistemic argument: “writing short-term policy goals into the Constitution . . . nearly always turn[s] out to have bad and unintended structural consequences,” in part because “amendments are passed piecemeal. The framers had to think about how the entire thing fit together.”309 I will claim that this argument commits a nirvana fallacy: it fails to compare amendments with the institutional alternatives for producing constitutional change, principally constitutional common law. The alternative to constitutional amendment is not a stable and epistemically impressive subconstitutional order; the alternative is continual judicial updating of the Constitution through flexible common-law constitutionalism. That practice is equally subject to the limits of reason and equally exposed to the risk of unintended consequences.
An Important Ambiguity There is a threshold ambiguity concerning the weight of the epistemic argument against constitutional amendment. Should it be understood to support invariable rejection of proposals for constitutional amendment, or a mere presumption against constitutional amendment? In what follows I shall take it as understood that the argument can be expressed in stronger or weaker versions. The tradeoff is that the argument becomes more plausible when implemented by a mere presumption, but the price for increased plausibility is diminished force. My main points are unaffected by the precise specification of the argument’s weight.
The Epistemic Argument Against Amendment The epistemic argument against constitutional amendment rests upon the limits of human foresight in the face of high costs of prediction and information-gathering. It is of course an admissible argument against a proposed amendment that the proposal will have some particular—identifiable and identified—bad consequences. The argument from bad unintended consequences, however, is different on two counts. First, the bad consequences are precisely those not foreseen310 at the time of debating and
309 310
The quotes in this paragraph are all taken from id. at 41–42. I will treat unforeseen and unintended as synonymous, ignoring the case of effects that are foreseen but not intended, as in the casuistic doctrine of “double effect.”
170
L L R
voting on the proposal. This gives the epistemic argument against amendment a generic quality. The difference is between a caution not to cross the street lest one be hit by that oncoming car, and a caution not to leave the house lest something bad occur. Second, the unforeseen bad consequences are sometimes said to be “structural.” Consider the following: [A] reason to resist writing short-term policy goals into the Constitution is that they nearly always turn out to have bad and unintended structural consequences. This is in part because amendments are passed piecemeal. In contrast, the Constitution was drafted as a whole at Philadelphia. The framers had to think about how the entire thing fit together.311 The argument in the passage above has the following components: (1) a concern with bad and unintended structural consequences, a concern that is underwritten by (2) a concern that amendments produce piecemeal and incoherent, as opposed to globally coherent, constitutional law, and by (3) an assumption that legal coherence is both feasible and desirable. I shall discuss each component in turn.
Coherence and Amendments The assumption embodied in (3) is questionable at best. There are institutional reasons to think that a degree of doctrinal incoherence is inevitable,312 and there are also good reasons to question the value of coherence, either in legislation313 or perhaps especially in constitutional law. Good coherence is better than incoherence, but bad coherence is worse than incoherence; coherence raises the stakes of constitutional decisionmaking by propagating either good or bad decisions through the legal system. There is also a dubious, but implicit, historical premise in the argument— that the framers of the United States Constitution designed a coherent scheme, as opposed to aggregating competing values and preferences, through horse-trading, into a patchwork document. Those issues aside, the rationale offered in (2) exemplifies the nirvana illusion. The comparison between the framers’ globally coherent design, 311 312 313
Sullivan, Constitutional Amendments, supra note 308, at 42. See Frank H. Easterbrook, Ways of Criticizing the Court, 95 Harv. L Rev. 802 (1982). See Andrei Marmor, Should We Value Legislative Integrity?, in The Least Examined Branch: The Role of Legislatures in the Constitutional State 125 (Richard W. Bauman and Tsvi Kahana eds., 2006).
U C C A
171
on the one hand, and piecemeal amendment, on the other, is not the right comparison to make. The principal substitute for formal amendment is not formal constitutional conventions, but judicial updating of constitutional law through flexible interpretation. The question, then, is whether piecemeal amendment produces greater incoherence than piecemeal judicial updating, carried out in particular litigated cases, by judicial institutions whose agenda is partly set by outside actors. There is little reason to believe that judicial updating under these conditions is more conducive to coherence than is the amendment process. We should disavow any implicit picture of judge-made constitutional law as an intricately crafted web of principles whose extension and weight have been reciprocally adjusted. Precisely because judicial updating requires overrulings, reinterpretations, and other breaks in the web of prior doctrine, a system that relies on judicial updating to supply constitutional change generates internal pressures towards incoherent doctrine. Constitutional adjudication in America has produced both Plessy v. Ferguson and Brown v. Board, both Lochner v. New York and West Coast Hotel v. Parrish, both Myers v. United States and Humphrey’s Executor, both Dennis v. United States and Brandenburg v. Ohio, both Wickard v. Filburn and Lopez v. United States, both Bowers v. Hardwick and Lawrence v. Texas. Whatever else can be said about this judicial work-product, and whatever other justifications can be given for judge-made constitutional law, deep inner coherence does not seem either a plausible description of the terrain or even a plausible regulative ideal for the system. The collective authorship of “Great and Extraordinary Occasions,” a set of “Guidelines for Constitutional Change,”314 advance a related argument from coherence. The argument is best illustrated with the example of the proposed amendment that would have stripped free speech protection from the act of burning the American flag. The guideline authors object that the amendment would have been “[in]consistent with related constitutional doctrine that the amendment leaves intact.”315 The problem arises “when framers of amendments focus narrowly on specific outcomes without also thinking more broadly about general legal principles.”316 The Court’s decision to label flag-burning protected speech was derived from established background principles of judge-made free speech law, while a 314
315 316
Great and Extraordinary Occasions: Guidelines for Constitutional Change 3 (1999) [hereinafter “Guidelines”], available at http://www.constitutionproject.org/pdf/ great_and_extraordinary_occasions.pdf. Id. at 17. Id. at 18.
172
L L R
flag desecration amendment would merely have overturned a particular judicial outcome. Here the nirvana illusion is twofold. The argument underestimates the capacity of specific positive enactments to generate broader principles in the future, and overestimates the extent to which background free speech doctrine rests on coherent principles.317 As to the first point, an argument emphasized by James Landis in the statutory setting, as against defenders of the common-law status quo, was that new statutes may themselves generate new principles when interpreted over time, in part by judges sympathetic to the aims of the new enactment.318 So too in the constitutional setting. It is easy to imagine future courts generalizing new principles from a flag desecration amendment, principles emphasizing the authority—the right, if you will—of enduring majorities to mark out as fundamental a limited class of national symbols or ideals or aspirations, and to grant those symbols immunity from the ordinary hurly-burly of free speech in an open society. The precise contour of such principles is not now apparent, but that is also true whenever courts embark on the development of new lines of constitutional doctrine. In both cases, coherentists should expect a process of mutual adjustment to occur, as new principles elbow their way into the constitutional arena and force old principles to reconcile themselves to a new, narrowed scope. As to the second point, free speech law cannot usefully be described as a coherent web of principles that yield particular decisions. The principles that do exist underdetermine the outcomes of many cases. After all, four justices thought that statutes banning flag-burning were consistent with the central free speech injunction of content neutrality, and it requires a great deal of confidence to declare them not only wrong, but trivially or obviously wrong. Thus much free speech law takes the form of rules, not principles—rules that in many cases have a narrow scope and highly specific content, and that are derived from background principles only with the help of supplemental empirical and institutional premises.319 Even if a flag desecration amendment generated no broader principles, it would not represent the intrusion of a particular outcome into a web of principle.
317
318 319
For a more thorough comparison of the collateral effects of amendments and statutes on constitutional precedents, see Frank Michelman, Saving Old Glory, 42 Stan. L. Rev. 1337 (1990). See James McCauley Landis, Statutes and Sources of Law, in Harvard Legal Essays 213 (1934). Frederick Schauer, The Boundaries of the First Amendment: A Preliminary Exploration of Constitutional Salience, 117 Harv. L. Rev. 1765 (2004).
U C C A
173
It would merely change the content of one or a few highly specific rules previously established by judges.
Bad Unintended Consequences and Judicial Updating Perhaps the concern with bad unintended consequences can be justified on other grounds even if the coherence rationale fails. An incoherent state of constitutional doctrine is just one type of bad unintended consequence that amendments might produce. Even if the particular concern is not that amendments will produce incoherent constitutional law, still we might hold a generic concern about the bad unforeseen (and possibly “structural”) consequences of constitutional amendment. Yet the nirvana illusion occurs on this more general level as well. It is wrong to believe that constitutional amendments represent risky action, while the steady-state of judge-made constitutional doctrine represents safe inaction. This is akin to the crudest defenses of the so-called “precautionary principle” in environmental law,320 or of the Hippocratic injunction to “do no harm.” Inaction may produce the medically worst outcome of all, and the status quo may itself contain dangerous environmental risks. Likewise, a persistent judicial refusal to update obsolete constitutional law can itself produce large political, social, and economic harms; this is a possible account of the mistakes of the Lochner Court.321 In a world of change, where constitutions must be updated somehow or other over time—if they are not to be replaced outright—the real alternative to formal amendment is not judicial “inaction,” but affirmative judicial action to update obsolete doctrine. Here the nirvana illusion is the failure to recognize that judicial updating, as a substitute for formal amendment, can itself produce bad and unforeseen structural consequences. The limits of reason bind the judges as well. I conclude that there is no way to leverage a concern for disrupting coherence, or a broader concern for unforeseen consequences more generally, into a generic epistemic concern about constitutional amendment. Anything that people or institutions do or fail to do may result in bad unforeseen consequences. Statutes may produce them, but so may the failure to enact statutory reforms; judicial decisions may produce them, but so may judicial “inaction”; and so too for constitutional amendments.
320 321
Cass R. Sunstein, Laws of Fear: Beyond the Precautionary Principle (2005). See Lochner v. New York, 198 U.S. 45 (1905).
174
L L R
The worry about the perverse consequences of amendment suggests nothing in particular.
A Note on Rights and Structure Sometimes, epistemic arguments do not target amendments as a class, but instead identify some subcategory of disfavored amendments. Usually such arguments turn on a distinction between “rights” provisions and “structural” provisions, and allow broader scope for amendments to affect structure than to affect rights. For concreteness, I will consider the following version of this view: All previous amendments can be categorized as either (1) expanding individual rights or (2) improving government structure. Amendments outside these two categories should be rejected or disfavored, on the Burkean ground that there must be an implicit epistemic logic to the patterns of the past.322 This view seems untenable, on both empirical and normative grounds. The historical premise that no amendments have limited individual rights is very dubious. A major purpose of the Thirteenth Amendment was to contract an individual right: the substantive due process right to own slaves, identified by the Court in Scott v. Sandford.323 Section 1 of the amendment reallocated the right to control the slaves’ labor from the slave owners to the slaves themselves, a classic taking from A to give to B—which is why the fourth section of the Fourteenth Amendment later specified that no compensation would be due to the slave owners, as might otherwise have been the case. But let us suppose the historical premise true. The hard question is what normative significance, if any, the historical pattern might have. Why should the categories into which previous amendments happened to fall be taken as exclusive? Why should not other types of amendments be added as circumstances change? In particular, we might imagine a category of amendments that (3) restrict judicially-identified individual rights in order to protect collectively-held symbols or values or aspirations. (An amendment eliminating free speech protection for flag-burning would be an example).
322 323
Guidelines, supra note 314; Cass R. Sunstein, President Versus Precedent: Bush’s Reckless Bid for an Amendment Defies an Oval Office Tradition, La Times, Feb. 26, 2004, at B13. 60 U.S. (19 How.) 393, 450–52 (1857).
U C C A
175
Is there an epistemic argument against (3)? Burkeans may believe or hope that there is some immanent logic underlying our amendment practices, a logic that might justify taking categories (1) and (2) as exhausting the set of permissible amendments. Another possibility, however, is that our ability to sort the previous twenty-seven amendments into categories (1) and (2) is sheer curve-fitting. It is an artifact of the small number of amendments, in turn an artifact both of the stringency of Article V’s supermajority requirements, and, in more recent decades, of an intellectually misguided presumption against amendment. The list of approved amendments turns out to be sensitive to small changes in the Article V rules. Had Article V, for example, required the approval of only a majority of states, the Constitution would contain, inter alia, the 1924 Child Labor amendment.324 That amendment would have been a category (3) amendment in spades: it was opposed not only on the predictable ground of federalism, but on the ground that (in the words of its chief opponent) it was “a highly socialistic measure—an assault against individual liberty,”325 presumably the joint liberty of parents and employers to bind children to labor contracts. I conclude that the epistemic argument against constitutional amendment fails. There is no basis for global or presumptive epistemic skepticism of the amendment process as a means of constitutional change. The standard attempts to identify epistemic deficiencies in the amendment process, rooted in the fear of bad unintended consequences and in the limits of reason more generally, fail to compare the institutional alternatives: they identify features that are common to both the amendment process and the alternative institutional process of constitutional common law.
Constitutional Updating and the Limits of Reason: An Institutional Comparison I turn now to constructive analysis, attempting a comparative epistemic evaluation of institutional processes. The Constitution must be updated over time; on epistemic grounds, which institution (or set or mix of institutions) should be entrusted with the task? I begin by clarifying the 324
325
See Stephen M. Griffin, The Nominee is . . . Article V, 12 Const. Comm. 171 (1995). The amendment was ratified by twenty-eight states. See Richard B. Bernstein and Jerome Agel, Amending America: If We Love the Constitution So Much, Why Do We Keep Trying to Change It? 180 (1993). David E. Kyvig, Explicit and Authentic Acts: Amending the U.S. Constitution, 1776–1995, 259 (1996).
176
L L R
questions, and then examine the epistemic strengths and weaknesses of the amendment process, on the one hand, and common-law constitutionalism, on the other. The aim is to sketch the empirical conditions under which one process or the other might prove epistemically superior.
Preliminaries I will take the current supermajoritarian amendment rules as fixed. I assume that constitutional rules and political constraints, even taken together, leave some freedom of action or play within the system. Political actors may choose to steer more or less constitutional change, and varying types of constitutional problems, through the amendment process, on the one hand, or the processes of constitutional common law, on the other. We are interested in the question which choice they should make, under what circumstances, given the limits of reason.
Common-Law Constitutionalism, Popular Constitutionalism, and Judge-made Law But we must be clear about what exactly the alternatives are. The formal Article V amendment process is a well-defined means of constitutional updating, but the alternative of common-law constitutionalism is blurrier. On one view, most constitutional change outside of Article V is initiated by nonjudicial actors, and perhaps completed by those actors as well. Judges often simply acquiesce in structural innovations produced by legislative and executive actors, innovations that are so large-scale as to change the constitutional order. Consider, for example, the large role played by nonjudicial actors in the rise of the administrative state and the development of presidential power in the twentieth century. In American legal theory, a large recent literature on “popular constitutionalism” underscores this point.326 Nothing in what I say here is incompatible with the claims of popular constitutionalism. Large-scale social movements that advance distinctively “constitutional” claims can bring about either formal Article V amendments or, as legal scholar Bruce Ackerman emphasizes,327 can 326
327
See, e.g., 1 Bruce Ackerman, We the People: Foundations (1991); Bruce Ackerman, We the People: Transformations (1998); Richard D. Parker, “Here, the People Rule” (1994); Larry D. Kramer, The Supreme Court, 2000 Term—Foreword: We the Court, 115 Harv. L. Rev. 4 (2001). Ackerman, id. at 266–94.
U C C A
177
instead channel their energies into presidential elections, judicial appointments, and constitutional change through the courts. I focus on the latter stages of this process: the choice between the formal amendment path and the path of common-law constitutionalism, where those are substitutes. Any large-scale movement for constitutional reform will usually have to follow one or the other of these paths, or both. If neither formal amendment nor judge-made constitutional change eventuates, one may justifiably be skeptical that the claimed constitutional movement was popular at all. And usually one or the other path is chosen; the amendment process and common-law constitutionalism are partial substitutes for each other. Although nothing in my thesis depends on it, I add that we should also be careful not to understate the leading role of judges in producing or at least ratifying non–Article V constitutional change. Here the distinction between constitutional rights and constitutional structure is useful. As far as rights are concerned, few deny the leading role of the federal judiciary in constitutional updating. “The U.S. Constitution contains broad definitions of rights, and the task of amending their definitions to reflect changes in the country’s economic, social, and political characteristics has been largely carried out by the Supreme Court.”328 As for structure, it is more plausible to say that large-scale change occurs outside the judiciary, that the judges simply get out of the way of it, and that the outcomes are the same as those that would have occurred had the judges simply gotten out of the business of constitutional review altogether. But getting out of the way, as the judges did after the New Deal, required capacious interpretation of the national government’s powers— which was itself just another type of judicial updating. It is treacherous to assume that structural innovations can achieve constitutional status through silent judicial acquiescence. Such innovations have sometimes developed over generations, and seemingly become entrenched, only to be rejected by the judges when the issue was squarely posed. Consider the legislative veto, which became a ubiquitous feature of federal law between World War I and 1983, when the Supreme Court invalidated all such provisions at a stroke.329 Many thought the legislative veto was an entrenched nonjudicial structural change, a fait accompli that the judges would have to ratify; but the judges didn’t think so.
328 329
Dennis C. Mueller, Constitutional Democracy 223 (1996). INS v. Chadha, 462 U.S. 919 (1983). The legislative veto lives on as a legal tool in congressional-executive interaction, but a legally unenforceable one. See Louis Fisher, The Legislative Veto: Invalidated, It Survives, 56 L. & Contemp. Probs. 273 (1993).
178
L L R
In short, structural changes developed by the political branches are always constitutionally insecure until embodied in a formal amendment or until the judges put an affirmative stamp of approval on them; judicial acquiescence in the sense of inaction leaves a penumbra of uncertainty and a residual risk of judicial invalidation. Despite the New Deal’s transformation of American public law, the absence of a New Deal Amendment enables the Court to flirt with retro-restrictive interpretations of the Commerce Clause.330
Institutions, Constitutional Change, and Epistemic Capacities Institutional personnel, structure, and norms affect the epistemic capacities of federal legislators, state legislators, and other actors in the amendment process, on the one hand, and the judicial actors who ultimately oversee common-law constitutionalism, on the other. We can say something, perhaps a great deal, about the epistemic quality of these institutional systems without knowing what the right answers ultimately are—just as we can surmise, in general, that trial by jury is epistemically superior to trial by battle, even if we do not know who exactly is or is not guilty or innocent. In general, I suggest that the Article V process takes a relatively abstract perspective, detached from the facts of particular cases, which can produce a valuable form of critical distance; is free from the epistemic influence of past judicial decisions; and incorporates a relatively wider range of professions and perspectives, which produces valuable cognitive and deliberative diversity. Some of these differences produce both costs and benefits that vary across different domains; neither process is better universally or in the abstract. After identifying the structural tradeoffs, I outline the empirical conditions under which either process shows to best advantage.
Abstraction and Information A structural constraint on common-law constitutionalism is that decisions about constitutional updating are made in the setting of particular, litigated cases and controversies. By virtue of the prohibition on advisory opinions in the Article III courts, judges deciding constitutional cases 330
See, e.g., United States v. Morrison, 529 U.S. 598 (2000); Unites States v. Lopez, 514 U.S. 549 (1995).
U C C A
179
have before them flesh-and-blood parties whose particular circumstances are often arresting or dramatic, and who often come to exemplify various claims for or against constitutional change. This focus on the particular case is sometimes said, and often assumed, to improve the quality of judicial decisions. The concrete setting of litigated cases, on this view, provides judges with information that legislators or amendment drafters lack. The contrast is just a matter of degree, of course. Participants in the amendment process may draw upon vivid anecdotes, while courts may, within limits, frame their deliberations in more or less abstract ways. But matters of degree are important. The case-or-controversy requirement in courts creates an insistent structural pressure towards focus on the concrete, a pressure that has no obvious analogue in the amendment process. There is a cost to the judicial focus on particulars.331 Abstraction may represent a virtue and detail a vice. It is not the case that decisions made with more information are always superior to decisions made with less information. For one thing, the information that comes from the presence of particular parties may produce a kind of inferential error: boundedly rational judges may err by assuming that the parties at bar exemplify the average or modal case within some larger class, whereas the parties may in fact represent atypical outliers in one or another respect. As Justice Byron White noted in dissent in INS v. Chadha, the Chadha majority invalidated “an entire class of statutes based on . . . a somewhat atypical and more-readily indictable exemplar of the class.”332 Another mechanism is salience, a heuristic that causes decisionmakers to overweight the importance of vivid, concrete foreground information and to underweight the importance of abstract, aggregated background information.333 The risk is that judicial updating will be distorted because judges overreact to the parties before them, perhaps by underestimating the relatively abstract social benefits that result from governmental infliction of vivid social harms on those parties. The amendment process is less tightly focused on highly salient particulars. Federal and state legislators considering amendments may draw upon the particulars of prior judicial decisions, but also possess a 331
332 333
For a more systematic statement of the arguments that follow, see Frederick Schauer, Do Cases Make Bad Law?, 73 U. Chi. L. Rev. 883 (2006). For responses that describe some advantages of common-law rulemaking, see Jeffrey J. Rachlinski, Bottom-Up Versus Top-Down Lawmaking, 73 U. Chi. L. Rev. 933 (2006); Emily Sherwin, Judges as Rulemakers, 73 U. Chi. L. Rev. 919 (2006). Chadha, 462 U.S. at 974 (White, J. dissenting). See Scott Plous, The Psychology of Judgment and Decision Making 125–26, 178–80 (1993).
180
L L R
broader range of data produced by legislative fact-finding and the submissions of competing interest groups. Although amici, expert witnesses and judicial notice of “legislative” facts allow the adjudicative process to partially compensate for the structurally superior fact-finding capacities of legislatures, common-law adjudication is in the end lashed to a particular set of facts in a way the amendment process is not. Most importantly, abstraction produces a kind of epistemic neutrality or impartiality. Participants in the amendment process enjoy a broader, systemic perspective, one that ranges simultaneously over sets of cases and that sees the costs in one part of the system that are necessary to produce benefits in another part. That enlarged perspective improves the ability of constitutional amenders, relative to that of constitutional judges, to make the large-scale tradeoffs, influenced by implicit or explicit costbenefit analysis, that are indispensable to systemic constitutional policymaking. The decentralized character of the judicial system and the necessity to consider “one case at a time” make it difficult for courts to take a systemic perspective.334 By contrast, the somewhat decentralized system of legislative committees, which hampers legislators from taking a systemic perspective on routine issues, has a smaller role to play where constitutional amendments are concerned.
Precedent and Path-Dependence Here we can return to some of the epistemic comparisons developed in chapters 1, 2, and 3. Common-law constitutionalists suggest that an important advantage of judge-made constitutional law is the discipline provided by precedent and by the related common-law emphasis on analogical reasoning. Precedent and analogical reasoning, it is said, conserve on the costs of information, encourage consistency over time, and ensure a kind of Burkean epistemic humility that dampens sudden or radical shifts in policy. These points, however, capture only one side of the ledger. Precedent, and the constraint that new decisions be related analogically to old decisions, effect a partial transfer of authority from today’s judges to yesterday’s judges. As against claims of ancestral wisdom, Bentham emphasized that prior generations necessarily possess less information than current generations, as discussed in Chapter 2. If the problem is that changing 334
See Donald L. Horowitz, The Courts and Social Policy (1977); Frank I. Michelman, Politics and Values or What’s Really Wrong with Rationality Review, 13 Creighton L. Rev. 487 (1979). Compare Cass R. Sunstein, One Case at a Time: Judicial Minimalism on the Supreme Court (1999).
U C C A
181
circumstances make constitutional updating necessary, it is not obvious why it is good that current judges should be bound either by the specific holdings or by the intellectual premises and assumptions of the past. Weak theories of precedent may build in an escape hatch for changed circumstances, but the escape hatch in turn weakens the whole structure, diluting the decisionmaking benefits said to flow from precedent. Another cost of precedent, discussed in Chapter 3, is path dependence. Because the existence of a precedent creates a positive cost to switching to a new legal rule, judges may be barred from reaching a rule that is optimal in the current circumstances, even though they would have reached it had the cases arisen in a different order. Precedent thus makes some optimal rules inaccessible to current decisionmakers. The constitutional rules that evolve through judicial updating may reach a local peak, but will be chronically incapable of reaching the global maximum. The institutions that participate in the process of formal amendment, principally federal and state legislatures, are less subject to these pathologies. Once the fixed costs of amendment have been paid, the drafters of constitutional amendments may draw upon society’s best current information, while ignoring precedential constraints that make it increasingly costly for courts to supply new rules as the needed changes become larger. The contrast is overdrawn, because—as Chapter 2 noted— legislatures use precedent in an informal way. But there is less likely to be relevant precedent when amendments rather than statutes are at issue, and precisely because the practice of legislative precedent is relatively less formalized than the practice of judicial precedent, legislative practice may capture most of the information benefits of precedent while minimizing its costs. Legislatures enacting an amendment are free to simply ignore precedent whenever legislators’ best current information clearly suggests that the constitutional rules should be changed.
Professionalism, Participation, and Epistemic Diversity We can also apply, in this setting, the points developed in Chapter 4 about the epistemic shortcomings of “Justices and Company”—shortcomings created by the Court’s inadequate professional diversity. Whatever else is true of common-law constitutional adjudication, it is clear that essentially all the direct participants in it are lawyers. Nonlawyers participate only indirectly, as expert witnesses or amici; their participation is always refracted through the agency of lawyers. The range of participating professions is far greater in the formal amendment process. Within the modern Congress, only 45 percent of legislators are lawyers (roughly), and a potpourri of
182
L L R
other backgrounds, skills, and experiences are represented. The principal comparative cost of common-law constitutionalism, in this respect, is not that it is elitist, but that it draws upon a remarkably narrow band of professional skills. On Jury Theorem grounds, the positive correlation of error in a group solely composed of lawyers reduces the overall quality of the Court’s performance in constitutional updating, even holding competence constant, as Chapter 4 detailed.
Tradeoffs, Variables, and Conditions Given structural tradeoffs between the Article V process and commonlaw constitutionalism, then in some domains the costs of one process will reach a zenith while the benefits reach a nadir. Here I offer a contextual appraisal of the relevant variables—of the conditions under which one process or the other shows to best advantage.
Public Judgments and Autarkic Law We have seen that lawyers’ expertise dominates the process of common law constitutionalism. That expertise appears in its best light where constitutional rules of a relatively technical character—what Chapter 4 called “autarkic” legal rules—must be adjusted over time. In settings of this sort the updating is itself relatively uncontroversial, although perhaps technically tricky; judges can accomplish the change at low cost; and it is not worth incurring the higher fixed cost of the formal amendment process, because it is unlikely that any disaffected group or large-scale social movement will attempt to unsettle the new, updated rules in future periods. An example involves one of the great success stories of nineteenthand twentieth-century constitutional adjudication, the so-called dormant or negative commerce clause. The problems in the area involve the reciprocal adjustment of competing state interests in a framework where all states benefit, in the long run, from national free trade; in which there is substantial consensus on the values to be promoted; and in which the judges have spent most of their time sorting out low-level questions about what rules can best be used to identify various de facto trade barriers. Many of these are technical lawyers’ questions par excellence—if state regulation imposes differential burdens on out-of-staters, should evidence of protectionist motive on the part of state officials be admitted?—and it is very hard to imagine that an amendment or series of amendments would have produced a better body of law, overall, than has the Court. The comparison is slightly inapposite here; dormant commerce law is also constitutional
U C C A
183
“common law” in the sense that the rules may be changed by statute as well as by amendment, and Congress has occasionally intervened. But the point is that in this area, judicial updating and development has been by far the dominant means of updating and has plausibly worked best of all. A very different problem arises when constitutional rules fall hopelessly out of step with large-scale changes in public judgments and values. It is not at all obvious that lawyers, as a professional class, enjoy any superior capacity to identify those values. Bentham’s insight, set out at the very beginning of the Introduction, was that a large group of legislators are systematically more likely to be correct about what “common reason” requires than a relatively small group of judges. Part of this is just the downside of lawyer’s technical expertise. Specialization and professionalization always introduces a kind of distortion, a narrowness or likemindedness in beliefs and commitments. Another part involves the overall epistemic costs of the institutional structures within which judges work. As Chapter 2 discussed, the judges’ political insulation gives them some distance from current electoral politics, which can produce a kind of epistemically advantageous detachment— what Alexander Bickel called “the ways of the scholar.”335 But judges pay a very large price for insulation, in the form of reduced information about what actual people desire and believe. What the electoral constraint does for federal and state legislators is to force them into closer contact with a broader range of views, professions, and social classes than most judges encounter. Where constitutional law has fallen out of step with large changes in public judgments and values, a legislature-centered process is epistemically superior.
Systemic Change Versus Piecemeal Change I have suggested that actors in the amendment process have relatively greater capacity to take a systemwide perspective. Where circumstances require structural change in the polity, the systemic perspective is at a premium, and the relative epistemic virtues of the amendment process are indispensable. It is unimaginable, in my view, that the basic readjustment of power between federal and state governments embodied in the Reconstruction amendments could have emerged from a process of common-law constitutionalism; nor would it have been desirable for it to do so. The polity-wide scale of the necessary changes, the inevitability of 335
Alexander M. Bickel, The Least Dangerous Branch 25 (1962).
184
L L R
interactions among those changes, and the risk that piecemeal change would create perverse effects, demanded that the issues be considered all at once, or at least in much larger decisional chunks than a case-by-case process of decisionmaking could provide. The Reconstruction Congresses saw a wide range of political actors, representing a wide range of skills, information, and interests, deliberate and bargain over large packages of rules simultaneously, reciprocally adjusting the choices made on different margins. The incremental character of common-law constitutionalism, by contrast, shows to best advantage where a series of piecemeal changes and small steps are in order. Consider the Supreme Court’s relatively successful efforts, moving cautiously, to update free speech and Fourth aAmendment privacy rules in light of twentieth-century changes in the technology of information, such as the widespread use of telephones. This sort of adjustment of rules and principles to slowly changing technical circumstances puts common-law constitutionalism in its best light. Under circumstances of this sort it will rarely be worth incurring the large up-front costs of the Article V process to make small adjustments. Conversely, the relative flexibility of the common-law process is a benefit in this sort of setting, where it is useful to be able to undo or reverse a failed adjustment at low cost in later periods.
Irreversible and Reversible Change Finally, common-law constitutionalism performs poorly where the benefits of rigid constitutional commitments are high, and the benefits of flexibility in future periods are low; and vice versa. Where it is desirable that new constitutional rules be tentative and reversible, because of a rapidly changing environment or because mistaken rules will have very high costs, common-law constitutionalism shows to advantage. Where it is desirable that new constitutional rules be costly to reverse, the amendment process is superior, just because amendments are much harder to destabilize in future periods. This is the case where, for example, the entrenchment of constitutional property rights serves as a costly signal that encourages investment,336 or where it is desirable to commit ex ante to political ground rules that will have important and contentious distributive effects when applied ex post. An example in the latter category involves the presidential succession rules, adjusted most recently by the Twentyfifth Amendment. 336
Daniel A. Farber, Rights as Signals, 31 J. Legal Stud. 83 (2002).
U C C A
185
Of course amendments will have to be interpreted by judges ex post as well, so the contrast is again just a matter of degree. But the need for judicial interpretation does not make amendments otiose or meaningless. Amendments can and often do sharply constrain interpretation within certain boundaries and can function as political and legal rallying points for groups and interests skirmishing in the courts in later periods. The need for later judicial interpretation gives amendments an added degree of flexibility, but as a comparative matter they are certainly less flexible (for good and for ill) than a common-law constitutional decision with identical content.
Interactions The proposal of chapters 2 and 3 that common-law constitutionalism should be entrusted primarily to legislatures has a substitution relationship with the proposal, in Chapter 4, for diversifying the Supreme Court’s professional composition. More of one requires less of the other, although a mix is likely to be optimal. How does the choice between amendment and judicial common-law constitutionalism interact with those proposals? The connections are complex. First, common-law constitutionalism effected through liquidating statutes, on the one hand, and through formal amendments, on the other, are also substitutes. As the former becomes more readily available, demand for the latter should decrease. However, the substitution is partial. Compared to liquidating statutes, formal amendments are both harder to enact and more durable when enacted, because harder to repeal. Legislatures have both tools at their disposal, and it is epistemically desirable that they should. The optimal mix of more and less durable forms of constitutional change is a difficult policy problem, and if the structure of legislative institutions makes them epistemically superior to courts, then legislatures will more frequently get the problem right. Second, diversifying the Court, by increasing its epistemic capacities, should lower the demand for either liquidating statutes or formal amendments—an extension of the substitution relationship discussed in Chapter 4. However, this does not imply that even in a world with a diversified Court, the optimal rate of either liquidating statutes or constitutional amendments is zero (the actual number of amendments over the past two generations, apart from the Twenty-seventh). The epistemic optimum is likely to involve a mix of reallocation to legislatures and diversification of the Court. While it is hard to be precise about where the optimum lies,
186
L L R
the one place it clearly does not lie is at the current extreme, where we have a Court with no lay justices, no constitutional amendments, and few liquidating statutesor at least few that the Court does not take it upon itself to invalidate.
Conclusion: Epistemic Legalism and Constitutional Change Epistemic legalism attempts to privilege judicial updating of constitutional law over formal amendment, by pointing to the limits of the enactors’ foresight. Amendments will generally produce unintended bad consequences, or so the claim runs. As against this, I suggest that the epistemic argument against constitutional amendment rests on a systematic nirvana fallacy, or failure of institutional comparison. The limits of reason constrain common-law constitutionalism executed by judges no less than the amendment process. Indeed, in a fair institutional comparison, structural and institutional features of the amendment process give it epistemic advantages over common-law constitutionalism, under an important range of identifiable conditions.
Conclusion
The Codified Constitution
U
nder the approach I have set out, what institutional arrangements would maximize law’s epistemic quality? What would constitutional law be like if the authority to make constitutional law were reallocated on the basis of the epistemic capacities of institutions, and if the institutions themselves were redesigned to improve epistemic performance? My basic claim is that given these goals, constitutional law can and should move from the common-law constitution to the codified constitution. This shift would have two principal components, one institutional, one legal. Institutionally, the power to develop and update constitutional law over time would be partially transferred from lawyer-judges on courts to nonlawyer-judges on courts and to nonlawyer non-judges in legislatures and elsewhere. Not all the power, of course. For one thing, lawyerjudges do not currently hold all the power to update constitutional law over time anyway; that is the valid core insight of the literature on “popular constitutionalism,” discussed in Chapter 5. For another, statutes would only receive deference from the judges if no clear mistake is present, under the Thayerian approach. Finally, once the epistemic argument against constitutional amendment is cleared away, the formal amendment process will be used more frequently than it is now; but as many have pointed out, 187
188
L L R
the high transaction costs of the Article V process will still exist and will still constrain the amount of formal amendment that can occur. Legally, constitutional law would increasingly take the form of positive enactments—both formal amendments and statutes. The latter would be ordinary law, from the standpoint of enactment, but would amount to a form of constitutional lawmaking by virtue of the legislature’s power to “liquidate” constitutional meaning, filling in constitutional gaps and ambiguities and adjusting constitutional arrangements to changing circumstances. Under the Thayerian approach, statutes would receive judicial deference whenever the Constitution is vague, general, aspirational, or ambiguous, which means most of the time. In this sense, statutes would make up most of the actual texture of constitutional law; the common-law constitution would just set minimal rules of play or broad outer boundaries, within which legislatures would be largely free to make and remake constitutional law over time. In short, my suggestion is that the final translation and culmination of the Benthamite project is to codify constitutional law itself. In America, constitutional law is partly codified today, but much of it is not. The U.S. federal Constitution is one of the shortest in the world, much shorter than in most states and other constitutional democracies. Most of American constitutional law is to be found “off the books”—or more precisely, not in the formal text of the Constitution. To the extent that our constitutional law is written at all, as opposed to taking the form of norms and practices, it is actually to be found not in “the constitutional text” but instead in the United States Reports, which contain the constitutional decisions of the Supreme Court. In one of the more puzzling effluvia of the common-law mind, however, these decisions are sometimes denominated “unwritten constitutional law,” by a flawed analogy to the genuinely unwritten constitutional law, which takes the form of constitutional conventions or customs or norms. Here is another example of the tendency to conflate judge-made law with customary law, a tendency we saw in both Hayek and Burke. Bentham himself developed a “Constitutional Code,” but it was basically unreadable, and in any event the background legal institutions he had in mind differed so fundamentally from those of the present—most notably, by virtue of the absence of anything like modern American-style judicial review, or even current British-style judicial review under the Human Rights Act 1998—that his code is hardly a plausible model. The codified constitution that I mean to suggest here is an updating of the Benthamite project. This version of the project would shift a great deal of
T C C
189
constitutional law out from the twilight world of the United States Reports, out from the domain of the common-law constitution, and into the domain of positive legal instruments, especially statutes. Not all constitutional law can or should be codified, of course. The experience of civil-law systems is constantly invoked to show that after bursts of codification, judge-made law grows interstitially like a kudzu vine, gradually overrunning the positive law, at least until a new burst of codification occurs. The point is not that this development is desirable, although many people who make the point also think that it is desirable, but rather that it is inevitable. Doubtless there is truth to this, but how much truth is the key question; it is all a matter of degree. Although there is no question that the constitutional law cannot be completely codified, there is also no question that far more of it could be codified than at present. To move from a 20/80 mix of constitutional code/constitutional common law to an 80/20 mix would be important, and pointing to the residual fraction of constitutional common law that is systemically inevitable does nothing to change that. I am not sure whether anyone really holds the view that because constitutional law cannot be fully codified, it cannot be codified at all, or any more than it happens to be now. If this view exists, however, it is what Bentham would have called a political fallacy,337 one whose effect, even if not intended as such, is to shore up the status quo of common-law constitutionalism against reform.
Codification and “Our Relative Ignorance of Fact” To a certain generation and stripe of legal theorists, the idea of a codified constitution will seem outlandish. On a standard view, everybody knows that codification simply cannot work for statutes, let alone for constitutions, principally because of the limits of reason. Legislative foresight is hampered by the fog of uncertainty that blankets the future; by scarce resources, especially time and an overloaded legislative agenda; and by the inevitability of changing circumstances. In the summa of this conventional wisdom, H.L.A. Hart wrote that legislators’ “inability to anticipate” dooms any attempt to “regulate, unambiguously and in advance, some sphere of conduct by means of general standards to be used without further official direction on particular occasions.” That inability arises from
337
See Jeremy Bentham, The Handbook of Political Fallacies (Harold Larrabee ed., 1952).
190
L L R
“a feature of the human predicament (and so the legislative one)”—“our relative ignorance of fact” and “our relative indeterminacy of aim.”338 This is true, but sadly noncomparative. Everything Hart says of legislators is also true of judges, who are interestingly absent from Hart’s parenthetical, which derives limits on legislative reason, but not judicial reason, from the human predicament. Unless judges are not human, the same logic goes through on both sides, barring some institutional difference. The limits of reason to which Hart points cut no ice where the utility of codification is concerned. Indeed, on the account I have given here, Hart’s view gets things backwards. The limits of reason matter most in a rapidly changing environment, where our “inability to anticipate” the future reaches a zenith. In that sort of environment, as discussed in Chapter 3, evolutionary models suggest that the relative advantages of legislation over common law increase. The problem is that common law is path-dependent; as environmental change becomes more rapid, the stickiness of common law becomes more costly. Legislation, even if imperfect, becomes increasingly superior by comparison, in part because it can adjust more rapidly to new conditions. Increasing statutorification is the best response to accelerating modernity. And that is what we have indeed observed, as the common law has over time increasingly given way to statutes. Common-law lawyers point out that civil-law systems have become more like common-law systems, as civil-law courts develop a more robust—even if officially disavowed— practice of precedent. The flip side of this coin, however, is that the “common law” systems have become so heavily statutory, due to the rise of the administrative state, that they have approached a state of piecemeal codification, however oxymoronic that sounds. The convergence of civillaw systems and common-law systems works both ways. And the convergence is to a thick system of statutory and regulatory law that dominates common-law precedent. At the level of constitutional theory, however, what we might call the Hartian nirvana fallacy—that the limits of foresight somehow afflict legislators alone, or more than other actors—is still alive and well; it has recently been made the cornerstone of another argument for a common-law approach to judicial review.339 It is past time for common-law constitutional theory to catch up with the increasing statutorification of common-law systems. The same process should extend to the common-law constitution. 338 339
H.L.A. Hart, The Concept of Law, 125–26 (1961). See W.J. Waluchow, A Common Law Theory of Judicial Review: The Living Tree (2007).
T C C
191
Durability and Epistemic Virtue Here a nonepistemic analogy may be useful. How durable are constitutions, in fact, and how durable should they be? What factors determine the lifespan of written constitutions? In an important recent work,340 it has been shown that ease of amendment increases the expected constitutional lifespan, as does the existence of constitutional review; that a commonlaw component in the legal system has no effect; and that “longer constitutions are more durable than shorter ones, suggesting that specificity matters.”341 This last finding will be counterintuitive to many American constitutional lawyers, who tend to hold a folk belief that general frameworks are more durable than specific ones. This work suggests, however, that “more detailed bargains indicate . . . relatively higher switching costs to produce a new bargain.”342 Overall, the counterintuitive conclusion on the score of durability fits hand-in-glove with the epistemic claim I advance here: “[C]onstitutions work best when they are most like ordinary statutes: relatively detailed and easy to modify.”343 That is exactly my suggestion, although on epistemic grounds rather than on the score of durability. Given the constraints of the Article V process, the codified constitution is the most that can be done to push constitutional law in the direction of greater specificity combined with the greater flexibility that the statutory and regulatory process provides. Here it is important to remember that the codified constitution includes not only formal amendments, at issue in Chapter 5, but also statutes that liquidate constitutional meaning, at issue in chapters 2 and 3. As an increasing share of constitutional law becomes statutory—as the space of constitutional law is occupied by statutes that, within the bounds of constitutional ambiguity, are allowed to define what the law is—then constitutional law becomes more code-like and yet also more flexible, because of the lower relative costs of changing statutes over time as the rate of environmental change increases.
A Note on Legal Origins For completeness, I must mention an important and interesting, though problematic, body of work on “legal origins.” This work claims, in essence, 340
341 342 343
Tom Ginsburg et al., The Lifespan of Written Constitutions (Law and Econ. Workshop Paper 3, 2008), available at http://repositories.cdlib.org/cgi/viewcontent.cgi?article=1212&context =berkeley_law_econ. Id. at 50 (emphasis in original). Id. at 51. Id. at 52.
192
L L R
that common-law systems outperform civil-law systems on various measures of economic success.344 To be clear, the work on legal origins is not directly on point here; it does not directly address the constitutional common law, as such, but rather focuses on the ordinary common law. However, its significance is analogical. A short extension might suggest that the common-law constitution has better economic consequences than the codified constitution. The problem is that as this body of work has developed, the core thesis has shifted. “Common-law” legal origin has left behind its ordinary meaning and become a general, and rather vague, synonym for freemarket ideology. In early versions of this work, legal origin denoted a difference between two large groups of legal families, common-law systems and civil-law systems (the latter having French, German, and Scandinavian variants). As critics pointed out, however, paradigmatic common-law nations like the U.K. and U.S. have statutorified much of their ordinary law as well;345 the administrative state has developed to the point where ordinary common law does not have many frontline responsibilities, even in the core domains of financial and corporate regulation on which the legal-origins literature focuses. This has caused the legal-origins proponents to clarify that by “common law” they mean “regulatory strategies [that] seek to sustain markets rather than replace them.”346 It is true that the common law is itself a system of regulation, but here the legal origins proponents are folding statute-based regulation by agencies into the definition of common law. This is a metaphorical use that does not undermine, and even supports, whatever other arguments for codification might exist. The general claim of the legal origins proponents is no longer about common law in the ordinary sense: “[T]o the extent that legal origins reflect fundamental approaches to social control of business, even legislation in common law countries would express the common law way of doing things.”347 This is to use common-law legal origin as a synonym for capitalism or freemarket ideology or low levels of regulation or something similar, a usage that rips it free of its original moorings.
344 345 346 347
For a recent overview, see Rafael La Porta et al., The Economic Consequences of Legal Origins (2007) (draft on file with author). See Mark J. Roe, Legal Origins, Politics, and Modern Stock Markets, 120 Harv. L. Rev. 462 (2006). See La Porta et al., supra note 174, at 43. Id. at 15.
T C C
193
Rationalism, the Limits of Reason, and the Codified Constitution Let me finish by relating the codified constitution to my largest suggestion, which runs throughout the foregoing chapters. The suggestion is that the limits of reason do not have the theoretical valence that so many supporters and critics of the common law have supposed. In a cartoonish version of this dialectic, celebrants of the common law point to the limits of reason to justify practices or presumptions that favor traditionalism, incrementalism, respect for custom and precedent, and appreciation of the law’s artificial reason. Critics of the common law, finding these arguments opaque, accuse their opponents of obscurantism and mystification.348 The common-lawyers, of either Burkean or Hayekian stripe, accuse their opponents of “excessive rationalism.” And so the interminable argument goes. My basic suggestion is that this debate rests on a false premise, one that gets things exactly backwards. The limits of reason do not support common-lawmaking in the first place, especially not in the constitutional domain; in that domain, the codified constitution is itself the best possible response to the limits of reason. To the extent that the critics of the common law have let common lawyers seize the high ground by appealing to the indisputable limits of human reason, that was an unnecessary concession; nor should critics of the common law have let themselves become the party of unbridled rationalism. Bentham’s proclivities certainly ran that way, but I have tried to show that other strands in his thought were fully compatible with a robust appreciation of the limits of reason. And where appropriate, I have tried to update and improve upon Bentham’s particular conclusions by incorporating that sort of appreciation, while remaining within the broad framework of his views. If Bentham’s project culminated in a Constitutional Code, so too a renewed respect for the limits of reason should culminate in a codified constitution. As I have emphasized throughout, this is a form of rationalism at one remove. It is a second-order rationalism that takes the limits of its own first-order capacities into account, and attempts to rationally evaluate and reform the legal system in light of those limits. As such, it is akin to Karl Popper’s “critical rationalism,” the chastened form of rationalism he deployed against Hayek’s claims about the limits of reason.349 I see no 348 349
See H.L.A. Hart, Bentham and the Demystification of the Law, 36 Mod. L. Rev. 2 (1978). See generally David W. Miller, Critical Rationalism: A Restatement and Defence (1994); Celia Lessa Kerstenetzky, Hayek and Popper on Ignorance and Intervention, 3 J. Institutional Econ. 33 (2007).
194
L L R
contradiction in such an enterprise; it is one of the main tasks of reason to tell us when reason runs out.350 What matters is that second-order rationalism be institutionally consistent, applying evenhandedly to legislatures, courts, agencies, and other actors. When the second-order analysis is complete and impartial, the limits of reason suggest a predominant role for legislative lawmaking, even in constitutional matters. Most ambitiously, the inevitable limits of human reason support a codified constitution.
350
On this theme, see generally Jon Elster, Ulysses and the Sirens: Studies in Rationality and Irrationality (1979).
Acknowledgments
I
wish, above all, to thank Emily, Spencer, Momma, Auntie Beeky, Grandpa and Umma for their love and patience during this project. I am also grateful to Dean Elena Kagan and many colleagues at Harvard Law School and elsewhere for their encouragement and support. Particular thanks (blame?) must go to Don Herzog, whose massive achievement, Poisoning the Minds of the Lower Orders, helped to inspire my Benthamite prejudices, and to Richard Fallon, who not only gave me insightful comments on most of the material but who also encouraged a book along these lines. Particular thanks also to Eric Posner, who commented helpfully on much of the foregoing material at one time or another. Cass Sunstein’s work, cited in the notes, influenced the book in innumerable ways, although our prejudices in this domain are opposed. Helpful comments and suggestions on the whole manuscript were offered by Herzog, John Manning, Matthew Stephenson, and an anonymous reviewer for the Oxford University Press. A superb conference on “La Sagesse Collective: Principes et Mecanismes (Collective Wisdom: Principles and Mechanisms),” organized by Jon Elster and Helene Landemore at the College de France on May 22–23, 2008, clarified some key points; thanks to Elster and the other participants for helpful comments on material that went into Chapter 1. Steven Horowitz, Joel Peters-Fransen and Elisabeth Theodore provided excellent research assistance, while Ellen Keng provided patient technical assistance and general good cheer. Parts of the foregoing chapters are heavily modified versions of earlier articles that appeared in various journals, as follows: ManyMinds Arguments in Legal Theory, Journal of Legal Analysis (forthcoming); Common-Law Constitutionalism and the Limits of Reason, 107 195
196
L L R
Columbia Law Review 1482 (2007); Should We Have Lay Justices?, 59 Stanford Law Review 1569 (2007); Constitutional Amendments and the Constitutional Common Law, in The Least Examined Branch: Legislatures in the Modern State (Richard W. Bauman and Tsvi Kahana eds., Cambridge University Press 2007). I thank the many colleagues and friends who offered comments on these papers.
Index
A accountants as Supreme Court members, 124, 142, 144, 150, 160–61 accounting cases, 137–39, 156 Ackerman, Bruce, 165n305, 176, 176nn326, 327 Adler, Matthew D., 113n205 advisory opinions, 178 Agel, Jerome, 175n324 agenda-setters, 25 Ali, S. Nageeb, 50n87 Alozie, Nicholas O., 124n241 amicus briefs institutional alternative component, 127 as source of expertise, 153 Amtsgericht’s lay judges, 128 analogical reasoning in common law, 180 analytical philosophy compared with social science tools, 22 ancestral wisdom, 180–81 antitrust cases, 156 Aristotle epistemic free-riding in deliberations, 47 one-many comparison, 53 Wisdom of the Multitude, 36–41 Arrow’s Theorem, 31 Article V. See constitutional amendments astrologer as Supreme Court justice, 160 Atkins v Virginia, 137 Auerbach, Jerold S., 140n276 Auriol, Emmanuelle, 84n153 Austria and lay judges, 128 autarkic cases, 134–35, 143–46, 150, 157–58, 182–83
B Babbitt, Bruce, 149 bankruptcy cases, 135 Battaglini, Marco, 50n88 Baum, Lawrence, 132nn258, 259 Becker, Gary S., 114n207 Bell, John, 129n252 Bentham, Jeremy ancestral wisdom argument’s weakness, 180–81 codified constitution and, 188 “common reason,” 1 counterfactual irrelevance of constitutional amendments, 187 epistemic bottlenecks in many-minds decisionmaking, 51 incentive effects on groups, 46 “judge and company,” 77–79, 124, 141 legal monopoly of adjudication, 124, 125 legislatures’ epistemic superiority, 11 simultaneous decisionmaking, 49 sociological substructure of law, 22–23, 162 Bernardo, Antonio E., 75n138, 148n291 Bernstein, Richard B., 175n324 biases constituents’, 87 epistemic superiority of legislatures, 90 lawyers’, 20, 126, 139–43, 140, 140n276 lay judges’, 137–39 in legislatures, 86–87 limits to reason, 1 See also correlated biases Biccieri, Christina, 74n134
197
198 Bickel, Alexander M., 183, 183n335 Bikhchandani, Sushil, 46n79, 74n130, 75n135, 148n289 biological evolution as efficient adaptation, 34–35, 98 Blackstone, Sir William, 18 Bentham’s critique of, 18 natural law constitutionalism in, 19 blind man and elephant metaphor 38 Bowers v Hardwick, 171 Bradley, Bill, 149 Brandenburg v Ohio, 171 Brennan, Geoffrey, 38n54, 39n57 Breyer, Justice Stephen, 123, 150 British Human Rights Act, 188 Brown v Board, 171 Burke, Edmund, 5–6, 15 basics of view, 62–63 Burkean paradox, 5–6, 75–77 Condorcetian interpretation of, 20, 60–96 Darwinian interpretation of, 60 dictum of, 58 epistemic legalism and, 2–3 evolution and efficiency, 104–05 Hayekian interpretation of, 60 judge-made law conflated with customary law, 188 many minds arguments, 10–11, 53 nonepistemic arguments, 9–10 Reflection on the Revolution in France, 58n94, 72n127 second-order rationality and, 64–6 business cases, 144, 156 C Calabresi, Steven G., 80nn145, 146 capital punishment commonality of questions in judicial review, 71 state re-enactments after court invalidation, 94 Caplan, Bryan, 87n157 Cardozo, Judge Benjamin balance among eccentricities of judges, 104, 116–17 Cardozo Theorem, 116–22 Carhart, Gonzales v, 71n123 Carhart, Stenberg v, 71nn122, 123 case or controversy requirement, 179 Casey, Planned Parenthood v, 126n247
I causal truth, 8 causation and constitutional amendments process, 186–88 central planning and Hayek, 102 Chevron deference, 152–53, 156 Chevron v NRDC, 152n294 Child Labor Amendment, 175 civil law systems, 189, 190, 191–94 Coase, Ronald H., 154n296 codified constitution, 186–96 defined, 4 cognitive diversity in Supreme Court, 125 cognitive heuristics and limits to reason, 1 coherence and constitutional amendments, 170–73 collecting information limits to reason, 1 See also information aggregation collective incoherence 39-41 collective wisdom and judicial precedent, 25 Commerce Clause, 178, 182–83 common law applicability to constitutional law, 61 Burke’s dictum, 58 cumulative epistemic legalism, 12–13 custom distorted, 54 distinguishing cases, 117–120 evolutionary account of, 5, 33–36, 98–102 “legal origins,” 101, 191–92 levels of rationalism, 3–4 limits of reason and, 193 many-minds arguments for, 26 nonepistemic claims, 9–10 overruling cases, 117–120 See also common-law constitutionalism; judicial lawmaking common-law constitutionalism abstraction and, 178–80 antonyms of, 13–15, 57–58 autarkic law and, 182–83 Benthamite critique translated into, 19, 47–48 Burkean paradox, 5–6 Burke’s dictum, 58 coherence of doctrine under, 171–72 common law distinguished, 15–16 concreteness of litigated cases, 178–80 Condorcetian argument for, 20
I constitutional amendments compared, 6–7 constitutional precedent, 79–80 cumulative epistemic competence, 95 defined, 2 denominator problem in many-minds arguments, 44 dormant commerce clause as example of, 182–83 epistemic and nonepistemic claims and tradeoffs, 9–10 epistemic costs and benefits of, 163–186 evolutionary interpretation of, 97–122 excessive rationalism charge, 3–4 incremental character of, 184 information aggregation and, 57–96 lay judges and, 161 legislative lawmaking as form of, 14, 61, 187, 194 limits of reason and, 193 liquidating statutes, 185 many-minds arguments for, 26 one-many arguments, 54–55 overruling’s effect, 120 polarization in, 120–22 reversible change, 184–85 supply-side models, 117 Thayerism and, 21, 93–96 See also common law; judicial lawmaking “common reason,” 1, 21 competence comparative epistemic competence of institutions, 3–4, 9, 53–54, 175–176, 187–96 cumulative epistemic competence of common-law constitutionalism, 95 dual competency of justices, 124, 127, 134, 149–51, 161 judges, 78, 91–92 many-minds arguments, 44–50 Condorcet, Marquis de, 19–21 Burke’s interpretation of, 60–96 selection effects in many minds decisionmaking, 45, 84 See also jury theorem Congress incentives to acquire information, 87–89 preference cycling, 40–41
199 professional diversity in, 84–85, 181–82 selection effects, 84 Congressional Research Service, 43 Connecticut, Palko v, 72n126 Conseil Constitutionnel’s lay judges, 128–29 constituents biases of, 87 electoral connection of legislators, 85, 86 constitutional amendments codified constitution and, 4, 187–96 common-law constitutionalism compared, 2, 14–15, 163–186 epistemic argument against, 168–70 epistemic costs and benefits of, 163–186 flag-burning, 171–73 intellectual constraints on process, 165–66 judicial interpretation of, 185 precedent’s use in, 180–81 reversible change, 184–85 “rights” provisions, 174–75, 177 structural constraints on process, 165–66 “structural” provisions, 174–75, 177 supposed futility of, 166–68 systemic change vs. piecemeal change, 183–84 unintended consequences and, 6–7, 163–186 constitutional amendments (cont.) weight of epistemic argument against, 169 constitutional law alternatives for updating, 164 correlated biases on, 87 demand-side model of efficiency, 112 efficiency as goal for, 112–13 See also common-law constitutionalism; constitutional amendments; constitutional precedent; constitutional texts; constitutional theory constitutional precedent Condorcetian interpretation of Burke and, 79–80 originalism and, 80–82 constitutional texts durability of, 191 epistemic virtue of, 191 size of, 188
200 constitutional texts (cont.) specificity in, 191 superiority to precedents, 60 constitutional theory Burke, Edmund, 5–6 correlated biases common professional training or background as source of, 145 judges, 78–79 lawyers, 20, 126, 139–43 legislatures, 86–87 self-selection as source of, 145 Supreme Court design and, 125, 139–43 cost of errors minimized with lay justices, 133–34, 143 costs of decisionmaking, 9 argument for lay justices, 133 Burkean judges and, 65 efficiency of legislation vs. litigation, 114–15 Couch, Carl J., 48n82 courts incentives to acquire information, 87–89 information cascades in, 74–75 legislative lawmaking deferred to, 60, 82, 122, 187–88 legislatures compared generally, 90 precedent as informational cascade, 76–77 selection effects in many minds decisionmaking, 46 statutes deferred to, 14, 82, 122, 187–88 See also lay judges; United States Supreme Court Covitz, Akiba, 140n292 cramdown issues, 135 criminal cases with lay judges, 128 Cross, Frank B., 114n210, 115n213 Cuba’s lay judges, 128–29 custom or tradition Bentham and, 18 Burke and, 10, 15, 59, 63–96 Burkean paradox and, 5 collective wisdom, 25 common-law constitutionalism and, 59–60 constitutional precedent, 80 cumulative epistemic legalism, 12
I denominator problem in many-minds arguments, 43 epistemic bottlenecks in many-minds decisionmaking, 51 epistemic superiority of, 15 Hayek and, 10, 15, 33, 35 as informational cascade, 76 information over time, 70 judicial lawmaking and, 2–3, 43 judicial precedent as, 43 Jury Theorem and, 66–68 nonlawyer Supreme Court Justices, 129–30 precedent distinguished, 68–69, 93 statutes’ superiority to, 60 superiority over legislation, 2 time and, 70 value of, 66 Czech Republic’s lay judges, 128 D Daicoff, Susan Swain, 140n276, 141n277 Darwin’s theory of natural selection, 34 Daugherty, Andrew F., 74n133, 75n136 decency. See “evolving standards of decency” deference to administrative lawmaking, 127, 152–53, 156 Chevron deference, 152–53, 156 intracourt deference in Supreme Court, 158–60 to juries, 127 to legislative lawmaking, 60, 82, 122, 187–88 to opinion leaders, 73n129 to special masters, 127 to statutes, 14, 82, 122, 187–88 Deitz-Uhler, Beth, 147n288 deliberation Aristotle’s doctrine, 36–41 independence and, 73 in Jury Theorem, 146–49 many-minds arguments and, 5, 25, 36–41, 146–49 Supreme Court, 146–49 demand-side models of common law evolution defined, 105 environmental change, rate of, 107–09
I fragility of assumptions, 106–07 internal problems, 106–09 invisible-hand mechanisms, 103–04 Dennis v United States, 171 denominators and many-minds arguments, 41–44 Denvir, John, 124nn242, 244 design of legal institutions, 6–7 Bentham, Jeremy, 18 limits of reason and, 2–4 dissent in groups, 75 distinguishing cases Cardozean model, 117–120 diversity epistemic superiority of legislatures, 11–12, 90 Jury Theorem, 29 See also professional diversity division of knowledge, 33–34, 102 doctors as Supreme Court members, 124, 145, 161 doctrinal paradox and deliberative model 40 dormant commerce clause and autarkic law, 182–83 dual competency of justices, 124, 127, 134, 149–51, 161 Due Process Clauses and efficiency goal, 113 Dulles, Trop v, 72n125, 126n248 Dunn, Thomas, 140n276 Dworkin, Ronald, 67n106, 112n204, 131nn254, 160n303, 255 E Easterbrook, Frank H., 170n312 ecology cases, 135–36 economics tools for design of institutions, 19 See also economists; law-and-economics movement economists dual competency of justices, 149–51, 161 Shleifer, Justice Andrei, 161 as Supreme Court members, 124, 140, 145, 149–51, 161 Edelman, 33n38 efficiency distinguishing cases, 117–120 evolution to, 98–116
201 Eichman, United States v, 71nn123 Eisgruber, Christopher L., 140n292 Eleventh Amendment, 119 Elhauge, Einer L., 107n188, 112n203, 114n210 Elliott, E. Donald, 152n294 Elster, Jon Bentham and Condorcet, 20n14, 46n77 evolution of social structures, 36n45, 100n169, 108n192 rationality, 194n350 Ely, John Hart, 141n278 emotions limits of reason, 1 many-minds arguments, 26, 47–48 energy cases, 156 environmental science cases, 135–36, 156, 161 epistemic bottlenecks denominator problem in many-minds arguments, 43 legislative process, 85 many-minds arguments and, 25, 50–53 epistemic claims defined, 7–8 tradeoffs, 9–10 epistemic free-riding, 46–47 sequential decisionmaking and, 49–50 Thayerism reduction of, 95–96 epistemic legalism components, 10–11 constitutional amendments and, 163–186 critique generally, 2–4 cumulative, 12–13 defined, 2–4, 7–8, 10–11 judicial lawmaking, 2–4 lay justices and, 162 many-minds arguments and, 26 social science underpinnings, 21–22 Supreme Court design and, 125 epistemic slack, 52, 53 Erlanger, Howard S., 140n276 error minimizing with lay justices, 133–34, 143 Estlund, David M., 37n49 Condorcet and Rousseau, 29n26, 30nn29 deference to opinion leaders, 30nn31, 73n129
202 Estlund, David M. (cont.) group deliberation, 30nn30, 38n51, 147n287, 159n302 ethnic diversity on Supreme Court, 124 eureka contributions of experts, 147 Everson, Stephen, 36n46 evolution Burke and, 59, 104–05 common-law constitutionalism interpretation, 97–122 demand-side models, 106–09 to efficiency, 98 to epistemic accuracy, 98 Hayek’s many-minds arguments, 33–36, 59 law-and-economics and, 102–04 mechanisms of, 105 path-dependence, 109–10 supply-side models, 116–17 “evolving standards of decency” capital punishment decisions, 72, 126, 137 common questions over time, 89 nonspecialized knowledge requirements, 139–42 executive decisionmakers and information aggregation, 90–91 Ex Parte Quirin, 136 expertise epistemic credential of, 20 expert witnesses, 153 incorporating legal expertise into Supreme Court, 154–56 lawyers’ professional expertise, 131, 182–83 lay judges, 137–39 marginal returns in group decisionmaking, 144–46 of Supreme Court justices, 146–56 expert witnesses, 153 F factual truth, 8 Fallon, Richard, 157n300 Farber, Daniel A., 184n336 Feddersen, Timothy J., 88n158 Federalist No. 78 Hamilton, Alexander, 17 Madison, James, 45, 48n81 Feld, Scott L., 147n286 Ferguson, Adam, 34n42
I Fernandez, Patricio A., 109nn194, 110nn196–198, 195 Fey, Mark, 28n20 Fisher, Louis, 177nn329 flag-burning common-law constitutionalism, 71 constitutional amendment, 171–73, 174 Fon, Vincy, 110n200, 111n201, 202 foreign affairs in Frankfurter canon, 94 Fourth Amendment, 184 France and lay judges, 128–29 Frankfurter, Justice Felix, 94 free market ideology and common law, 192 free speech flag-burning amendment and, 171–73, 174 incremental change to rights, 184 free trade and dormant commerce clause, 182–83 Fried, Charles, 125n246 Fukui, Yoshitaka, 74n134 G Garcia v San Antonio Metropolitan Transit Authority, 120 Gary-Bobo, Robert J., 84n153 Gaus, Gerald F., 100n171 gender diversity on Supreme Court, 124 Gennaoili, Nicola, 103–04 and Cardozo Theorem, 116–17, 119n225, 120 Germany’s lay judges, 128 Gigone, Daniel, 30n32 Ginsburg, Tom, 191nn341–43 Gonzales v Carhart, 71nn123 Goodin, Robert E., 29n25, 38n54, 83n152 Gould, Stephen Jay, 105n181 Granfield, Robert, 140n276 Green, Robert A., 159n301 Griffin, Stephen, 175n324 Grofman, Bernard, 29n24, 44n72, 147n286 groupthink lawyers and, 23 nonlawyer judges’ contribution to reducing, 126 professional diversity and, 11–12 Supreme Court, 134 in Supreme Court deliberations, 146–49 Guidelines for Constitutional Change, 171, 171nn314–16, 174n322
I H Hadfield, Gillian K., 106n185 Hamburger, Phillip A., 58n95 Hamdan v Rumsfeld, 136, 137n270 Hamilton, Alexander constitutional perspective, 21 Federalist 78, 17 Harsanyi, John C., 81n148 Hart, H.L.A., 189–90, 190n338, 193n348 Hasnas, John, 43n70, 99n168 Hastie, Reid, 30n32 Hathaway, Oona A., 109n193 Hayek, F.A. von division of knowledge, 33–34 epistemic legalism and, 2–3 evolutionary account of common law, 5, 33–36, 98–102, 188 Karl Popper’s critique of, 193 many minds arguments, 10–11, 53 nonepistemic arguments, 9–10 health cases, 136 Hedegard, James M., 140n276 herding in Jury Theorem, 147 in Supreme Court deliberations, 147–49 Herzog, Don, 100n170 hierarchical organization of executive branch, 91 Hippocratic injunction, 173 Hirsch, Adam J., 106n185 historians dual competency of Supreme Court Justices, 150, 161 as Supreme Court members, 123, 124, 145, 150, 161 historical analysis in cases, 136–37 Holmes, Oliver Wendell, 71nn124 Horowitz, Donald L., 180n334 House of Lords’ lay judges, 128 Humphrey’s Executor, 171 Hylton, Keith N., 107n191 I Ideological diversity in Supreme Court, 124 incentives’ effects on groups, 46–47, 75 incoherence, collective, 39–41 independence of groups members’ views, 29–30, 73–75
203 information aggregation common-law constitutionalism and, 57–96 epistemic superiority of legislatures, 11 many-minds arguments, 5, 25 supply-side models, 116–17 information cascades Burkean paradox, 5–6, 75–77 independence of group members violated, 73–75 rational ignorance and, 46 sequential decisionmaking and, 50 in Supreme Court deliberations, 147 institutional evolution, 25 many-minds arguments and, 5 institutions alternatives for constitutional updating, 163–65, 175–176, 187–96 Bentham’s institutional commitments, 17–21 cognitive diversity in design of, 125 comparative epistemic competence of, 3–4, 9, 53–54, 175–176, 187–96 constitutional precedent, 79–80 epistemic bottlenecks in, 25 epistemic capacities of, 178 evolutionary analysis of, 25 information aggregation, 25 professional diversity, 84–85, 181–82 See also administrative agencies; congress; courts; design of institutions; legislatures INS v Chadha, 177nn329, 179 intentional mechanisms common law benefits, 99 Hayek’s many-minds arguments, 34–36 invisible-hand mechanisms Burke, and, 5, 97 common law benefits, 99 Hayek’s many-minds arguments, 34–36 law and economics models, 103–04 iron law of oligarchy epistemic agenda-setters, 27, 53, 54 epistemic bottlenecks and, 53 Italy’s lay judges, 128–29 J Japan’s lay judges, 128 Johnson, Texas v, 71nn121, 123
204 judges motivation of, 131–32 political insulation of, 183 See also courts; lay judges; United States Supreme Court judgment’s locus in courts, 17, 21 judicial lawmaking Burkean view, 62–63 cumulative epistemic legalism, 12–13 denominator problem in many-minds arguments, 43 epistemic bottlenecks in many-minds decisionmaking, 51 epistemic legalism, 2–4 legislative lawmaking compared, 6, 82–87 many-minds arguments for, 26 rules, 110–11 sociological substructure of law, 22–23 standards, 110–11 unintended consequences of, 7 See also common law; common-law constitutionalism judicial motivation, 131–32 judiciary committees, 84 specialization’s effects, 85 juries as object of deference, 127 Jury Theorem, 19–21 Burke and, 5 Burke’s dictum and, 58–59 common deliberation or discussion, 30 common-law constitutionalism and, 60–96 common questions, 71–73 deliberation and, 37, 146 diversity, 29 epistemic free-riding, 46–47 evolutionary many-minds arguments and, 34 herding, 73–75, 147 independence of groups members’ views, 29–30 informational interpretation of, 5, 66, 68 lay justices and, 138, 141 many-minds arguments, 5, 28–33 one-many comparison, 53 polling model, 31–32, 68, 71n118 precedent and, 66–68 preference aggregation, 31
I professional diversity among Supreme Court Justices and, 125–26, 182 rational ignorance, 46 right answer, 21 tradition and, 66–68 whose minds, 41–42 worse minds and, 44 justices. See United States Supreme Court K Kaldor-Hicks efficiency, 102–03, 113, 116 Kaplow, Louis, 110nn199, 113n206, 200 Karotkin, D., 46n78 Kay, Susan, 140n276 Kerstenetzky, Celia Lessa, 102n177, 193n349 Kitahara, Minoru, 47n80 Klegon, Douglas A., 140n276 Komesar, Neil, 86n156 Kosovo’s lay judges, 128–29 Kramer, Larry D., 176n326 Kraus, Chadd K., 84n154 Kronman, Anthony, 63n97 Kyvig, David E., 175n325 L Ladha, Krishna K., 30n28, 42n69, 141nn278–82 Lancaster, Kelvin, 101n175 Landes, William M., 106n186 Landgericht’s lay judges, 128 Landis, James McCauley, 172, 172n318 Landwehr, Lawrence J., 140n276 Lanni, Adriaan, 37n47 La Porta, Rafael, 101n174, 192nn344, 346, 347 Latvia’s lay judges, 128–29 law-and-economics movement, 17, 18, 99 efficiency in, 102–04 information in, 102–04 law clerks as source of expertise, 153–54 Lawrence v Texas, 171 lawyers correlated biases of, 20, 126, 139–43 members of court, 23 professional biases of, 140–41, 140n276 professional expertise of, 131, 182–83 self-interest of, 22–23 self-selection by, 20, 141, 145, 151 systemic biases of, 141
I lay judges bias of, 137–39 comparison with legislative lawmaking, 161 epistemic legalism and, 162 expertise of, 137–39 members of courts, 20, 23, 128–30 number of, 144–46 as Supreme Court Justices, 123–162 leaders as epistemic bottlenecks in institutions, 25 LeBon, Gustave, 48n82 “legal origins,” 101, 107, 191–92 legal truth, 8 legislative lawmaking, 15–16 advantages of, 4 antonym of common-law constitutionalism, 14 codified constitution, 4, 190 deference to, 127, 152–53, 156 delegation to executive and administrative agencies, 15–16 efficiency of, 113–15 epistemic bottlenecks in many-minds decisionmaking, 51–52 epistemic superiority of, 11–12, 161, 190 judicial lawmaking compared, 2–4, 6 judicial review of, 12–13, 82 many-minds arguments for, 10, 26 numbers and selection effects, 83–4 precedents, 13 professional diversity, 84–85 rate of change as indicator of superiority of, 92 rise of, 176, 190 rules, 110–11 standards, 110–11 superiority to precedents, 60 unintended consequences, 10 Waldron, Jeremy, 22 legislative veto, 177 legislators abstention from voting, 88 incentives to acquire information, 87–89 legislatures bias in, 86–87 committee systems, 85–86 courts compared generally, 90 epistemic superiority of, 90 numbers, 83–84
205 organization, 85–87 precedent-based action, 89–90 professional politicians, 86 selection effects, 83–84 veto power, 177 See also Congress; legislative lawmaking; legislators levels of rationalism among institutions, 3–4 Levine, Felicia J., 140n276 limits to reason, 1 Lipsey, R.G., 101n175 List, Christian, 29n25, 83n152 Lithwick, Dahlia, 123 litigants as source of expertise, 153 Llewellyn, Karl N., 119n223 Lochner v New York, 171, 173n321 Lopez-de-Silanes, Florencio, 101n174 Lopez v United States, 171 Lutz, Donald S., 165n304 M Machiavelli and leadership, 52 Mack, Eric, 102n176 Madison, James emotional effects in many-minds decisionmaking, 48 epistemic bottlenecks in many-minds decisionmaking, 50 selection effects in many-minds decisionmaking, 45 Mahoney, Paul G., 101n174 majority rule and many-minds arguments, 41 many-minds arguments, 5, 25–55 Aristotle and deliberation, 28, 36–41, 47 common-law constitutionalism, 6 comparative epistemic competence, 53–54 competence of minds, 44–50 component of epistemic legalism, 10–11 Condorcet and information aggregation, 28–33 deliberation, 28 denominators, 41–44 emotion effects, 47–48 epistemic bottlenecks, 50–53 evolution, 28 filters, 41–53 Hayek and evolution, 28, 33–36
206 many-minds arguments (cont.) incentive effects, 46–47 information aggregation, 27–33 judgment aggregation, 32 Jury Theorem, 28–33, 41–42 limits of, 54–55 majority rule and, 41 multiplicands, 41–44 preference aggregation, 32 selection effects, 45–46 sequential decisionmaking, 48–50 simultaneous decisionmaking, 48–50 social influences, 47–48 taxonomy, 27–28 whose minds, 26, 41–44 worse minds, 26, 44–50 marginal returns in group decisionmaking, 144–46 markets coordination of individual expectations, 100 Hayek’s analogies, 98 Hayek’s many-minds arguments, 33–36 Smith, Adam, 33, 34–35 Marmor, Andrei, 170n313 Martinelli, Cesar, 88n159 medical science cases, 136, 142, 144–45 dual competency of justices, 149 Merrill, Thomas W., 63nn96, 66n103, 98, 115n212 methodological diversity on Supreme Court, 124 Miccosukee Tribe of Indians, South Florida Water Management District v, 135 Michaels, Robert, 53n92 Michelman, Frank I., 172n317, 180n334 Miles, Thomas J., 83n151 military cases, 136, 142, 161 Miller, David W., 193n349 Miller, Gary, 30n28, 42n69 Miller, Nicholas R., 32n37 Mitchell, George, 149 Montesquieu and selection effects in many minds decisionmaking, 46 Moore, Michael, 76n139 moral theorizing antonym of common-law constitutionalism, 14 efficiency’s status, 113
I moral truth, 8 Mueller, Dennis C., 177nn328 multiplicands and many-minds arguments, 41–44 Myers v United States, 171 N National League of Cities v Usery, 120 natural law, 18 natural selection theory of Darwin, 34 negative commerce clause, 182–83 New Deal, 177, 178 nonautarkic cases, 134–35 nonexperts’ epistemic credential, 20 nonlawyers. See lay judges numerosity average competence and, 44 emotions in decisionmaking bodies, 48 epistemic superiority of legislatures, 11–12, 90 lay judges, 144–46, 161 See also many-minds arguments O objective body of knowledge assumed for lay justices argument, 126, 130 oligarchy. See iron law of oligarchy Olson, Mancur, 114n209 oral arguments as institutional alternative component, 127 originalism, 80–82 antonym of common-law constitutionalism, 13 criticism of, 92 epistemic failings of, 81–82 Orth, John V., 129n253 overruling cases in Cardozean model, 117–120 P Page, Scott, 27, 29nn23, 37n48, 125n245 Palko v Connecticut, 72n126 Parisi, Francesco, 110n200, 111n201, 202 Parker, Richard D., 176n326 Parliament lay judges, 128 numbers and selection effects, 84 Paroush, J., 46n78
I partial-birth abortion and common-law constitutionalism, 71 path-dependence of common law, 109–10, 181 Pesendorfer, Wolfgang, 88n158 philospher-kings, 160–61 Pitts, Patrice M., 124n241 Planned Parenthood v Casey, 126n247 Plessy v Ferguson, 171 Plous, Scott, 179n333 polarization in constitutional cases, 117–22 in deliberating groups, 147 political ground rules, 184 political philosophy antonym of common-law constitutionalism, 14 social science tools compared, 22 political science, 19 politics and dual competency of justices, 149–50 Ponzetto, Giacomo A.M., 109nn194, 110nn196–198, 195 Popper, Karl, 193 Posner, Eric A., 42nn67, 68, 113n205 Posner, Judge Richard A. dual competency in law and economics, 149 information sources for Supreme Court, 153n296 intentionalist argument for common law benefits, 99, 103 law-and-economics movement, 17, 18 methodological diversity of Supreme Court, 124n241 power causing failure of independence, 73–74 precautionary principle, 173 precedent Burke and, 63–96 in civil law courts, 190 commonality of questions over time, 71–73 common-law constitutionalism and, 59, 180–81 denominator problem in many-minds arguments, 43 distinguishing cases, 117–120 epistemic bottlenecks in many-minds decisionmaking, 51
207 epistemic superiority of legislatures, 90 as informational cascade, 76 Jury Theorem and, 66–68 legislative lawmaking, 13 legislative use of, 181 many-minds arguments, 25 one-many comparison, 54 as sequential decisionmaking, 49 statutes’ superiority to, 60 tradition distinguished, 68–69, 93 value of, 66 See also common law; common-law constitutionalism precedential cascades, 74–75 prediction markets and Hayek, 35 preference aggregation model, 32 preference cycling, 40–41 presidential power in twentieth century, 176 presidential succession rules, 184 price system argument of Hayek, 102 Printz v United States, 120n232 Pritchard, A.C., 43n71, 114n211 privacy rules and incremental change, 184 professional diversity in Congress, 84–85 nonlawyer judges as source of, 16, 23, 186 on Supreme Court, 85, 124, 181–82, 185–86 professional politicians, 86 Q Quirin, Ex Parte, 136 R Rachlinski, Jeffrey J., 179n331 racial diversity on Supreme Court, 124 rate of change common law’s evolution toward efficiency, 100, 111–12 legislative lawmaking’s superiority, 92, 109–10, 115–16 rational ignorance endogenous information and, 87–89 incentive effects on groups, 46 Reconstruction Amendments, 183–84, 187 Rehnquist, William H., 119n224 Reinganum, Jennifer F., 74n133, 75n136 rent-seeking litigation vs lobbying, 107
208 representativeness of legislatures, 11–12 reputational cascades in Supreme Court deliberations, 147 Rheingold, Howard, 28n22 right answers assumptions, 67 autarkic cases, 134–35 evolutionary interpretation of commonlaw constitutionalism, 97 Jury Theorem and, 31 nonautarkic cases, 134–35 Supreme Court and, 125, 126, 131 what are, 21 Roe, Mark J., 192nn345 Roosevelt Administration, 137 Roper v Simmons, 126n248, 137 Rousseau common interest and, 29 Condorcet and, 29n26, 30nn29 Rubin, Paul H., 99n167, 106nn183, 107nn187, 114n208, 184, 188 rulemaking by Supreme Court, 157–58 rules distinguished from standards, 110–11 Rumsfeld, Hamdan v, 136, 137n270 S salience in constitutional updating, 179 San Antonio Metropolitan Transit Authority, Garcia v, 120 Sanders, Anthony B., 100n172 Saxe, John Godfrey, 38n53 Scalia, Justice Antonin, 123 Schauer, Fredrick caselaw, 179n331 free speech law, 172n319 lay justices, 124nn242, 131n256, 243 motivation of judges, 132n259 Schofield, Norman, 30n28 Scott v Sandford, 174 SCS Credit Corp., Till v, 135 search and seizure rules, 184 second best theory, 101–02 second-order rationality Burke and, 64–66 codified constitution and, 193–94 Segal, Jeffrey A., 132n258 Sekiguchi, Yohei, 47n80 self-interest limits to reason, 1
I sociological substructure of law, 22–23, 77–79 self-selection among professional classes, 20, 141, 145, 151 separation of powers and Frankfurter, 94 sequential decisionmaking in many-minds arguments, 48–50 Shavell, Steven M., 113n206 Sherwin, Emily, 179n331 Shleifer, Andrei Cardozo Theorem, 116–17, 119n225, 120 “legal origins,” 101n174 model of common law’s efficiency, 103–04 as Supreme Court justices, 161 Simmons, Roper v, 126n248, 137 simultaneous decisionmaking in manyminds arguments, 48–50 sincerity and free-riding, 47 slavery, 174, 187 “smart mobs” and many-minds arguments, 28 Smith, Adam, 33 social coordination, 102 social influences and many-minds arguments, 26, 47–48 social psychology and wisdom of crowds, 30 social science Bentham updated with, 19, 58 support for lay members of Supreme Court, 125 underpinnings for epistemic capacities of legislatures, 21–22 social wealth, 99, 103 social will’s location in legislature, 16–17, 21 soldiers as Supreme Court members, 124, 145, 161 Solum, Lawrence B., 67n107 South Florida Water Management District v Miccosukee Tribe of Indians, 135 sovereign immunity and polarization in common-law constitutionalism, 120 Spaeth, Harold L., 132n258 special masters deference to, 127 as source of expertise, 153–54
I spontaneous order and Hayek’s many-minds arguments, 34–36 standards distinguished from rules, 110–11 Starr, Kenneth W., 156n299 Stasser, Garry, 147n288 state sovereign immunity and polarization in common-law constitutionalism, 120 statutes codified constitution and, 4, 187–96 commonality of questions in judicial review, 71 interpretation of, 16 invalidation by courts, 12–13 judicial deference to, 14, 82, 122, 187–88 liquidating constitutional meaning, 185, 191 path-dependence and, 109–10 superiority to precedents, 60 “wisdom of the cradle,” 81–82 Stenberg v Carhart, 71nn122, 123 Stone, Julius, 119n223 Strauss, David A. common-law constitutionalism, 63n96, 64n99, 65nn100, 66n103, 80n147, 101 constitutional amendments’ irrelevance, 166–68 evolutionary interpretation of Burke, 97n164 ideological diversity of Supreme Court, 124n241, 155n298 precedent distinguished from tradition, 68n109 Suarez, Thomas A., 84n154 Sugden, Robert, 35n44, 101n173 Sullivan, Kathleen M., 110n199, 168n308, 169n309, 170n311 Sunstein, Cass R. Burkean as sword or shield, 92n160 Burkean Minimalism, 20n12 Condorcetian interpretation of Burke, 39n56, 63n96, 66nn102–4, 67n105 dissent in groups, 75n137 evolutionary interpretation of Burke, 97n164 experts’ contributions to group, 147n288, 148n290 extremes in group decisionmaking, 37n50 foreign law considered by American courts, 42nn 67, 68
209 ideological diversity of Supreme Court, 124n241, 155n298 judicial minimalism, 180n334 many-minds arguments, 5 moral philosophy, 160n303 precautionary principle, 173n320 sincerity of voters, 83n151 supermajority rules for constitutional amendments, 166, 175, 176 Jury Theorem and, 28n20 supply-side models Cardozo Theorem, 116–17 invisible-hand mechanisms, 103–04 supreme court. See United States Supreme Court Surowiecki, James, 8n5, 20n13 “wisdom of crowds,” 28n21 Sweden’s lay judges, 128–29 T Takings Clause, 113 Talley, Eric, 74n132 Tapp, June Louin, 140n276 tax cases, 137–38, 156 Texas v Johnson, 71nn121, 123 Thayer, James Bradley, 16–21, 82n150 information aggregation and, 93–96 judicial deference to legislatures, 82, 122, 187 Thayerism, 21, 93–96 Thirteenth Amendment, 174, 187 Till v SCS Credit Corp., 135 tradition. See custom or tradition Tragedy of the Commons, 35 transcendent dimensions of liberty, 139–42 Tribe, Laurence H., 119n227 Trop v Dulles, 72n125, 126n248 truth assumptions, 67 defined, 7–8 wisdom compared, 8–9 Tushnet, Mark, 140n292 Twenty-fifth Amendment, 184 Twenty-seventh Amendment, 166 U Udana, 38n52 Ullmann-Margalit, Edna, 34nn41, 42
210 unforeseen consequences. See Unintended consequences unintended consequences bad unintended consequences, 173–74 common-law constitutionalism and, 6–7, 163–186 component of epistemic legalism, 10–11 constitutional amendments and, 163–186 unforeseen consequences as synonymous, 169n310 United States, Printz v, 120n232 United States Reports, 188–89 United States Supreme Court agenda-setting in, 156, 157–58 autarkic cases, 134–35 cognitive diversity of, 125 deliberation by, 134, 146–49 denominator problem in many-minds arguments, 44 dual competency of Justices, 124, 127, 134, 149–51, 161 ecology cases, 135–36 economics cases, 135 endogenous agendas and rules, 157–58 environmental science cases, 135–36 error-minimization with lay justices, 133–34, 143 ethnic diversity of, 124 expertise in, 146–49, 150–56 finance cases, 135 Frankfurter canon, 94 gender diversity of, 124 groupthink in, 134, 146–49 health cases, 136 historical analysis in cases, 136–37 ideological diversity of, 124 incentives to acquire information, 88 institutional alternatives to, 127 intracourt deference in, 158–60 lay members of, 123–162 many-minds arguments for professional diversity, 125–26, 134 medical science cases, 136 methodological diversity of, 124 military cases, 136 motivation of justices, 131–32 nonlawyer members, 16, 23, 123–162, 186
I nonspecialized knowledge requirements, 137 one-many arguments, 54–55 philospher-kings, 160–61 precedent and, 69–70 professional diversity of, 85, 124, 181–82, 185–86 racial diversity of, 124 representation of expertise, 150–56 sequential decisionmaking, 49 specialized knowledge requirements, 126–27, 137 tradition distinguished from precedent, 69 Warren Court, 69, 187 United States v Eichman, 71nn123 United States v Lopez, 178nn330 United States v Morrison, 178nn330 United States v Virginia, 153 Usery, National League of Cities v, 120 V Vinson, Linda, 124n241 Virginia, Atkins v, 137 Virginia, United States v, 153 W Waldron, Jeremy, 22, 29n26, 30n29, 67n108 common questions in Jury Theorem, 71n119 independence in Jury Theorem, 73n128 moral philosophy as objective knowledge, 160n303 Waluchow, W.J., 190n339 Warren Court, 69, 187 water management cases, 135–36 Welch, Ivo, 148n291 West Coast Hotel v Parrish, 171 Wickard v Filburn, 171 will. See social will Willging, Thomas, 140n276 Williamson, Oliver E., 152n293 Wilson, Chris, 53n93 Wilson, Dennis B., 139nn274, 275 wisdom ancestral, 180–81 truth compared, 8–9 See also “wisdom of crowds”
I “wisdom of crowds” Aristotle and, 36–41 many-minds arguments, 5, 28 truth compared, 8 wisdom of experts compared, 20 “wisdom of the cradle,” 81–2 Y Young, Ernest, 63n96, 66n103 evolutionary interpretation of Burke, 97n164
211 Z Zywicki, Todd J. competition among legal institutions, 105n182 costs of legislative change, 115n212 custom and common law, 99n168, 107n190 market coordination of individual expectations, 100n172 rent-seeking, 114n211 Supreme Court as social planners, 43n72